Beruflich Dokumente
Kultur Dokumente
Bert Dufrasne
Roger Eriksson
Andrew Greenfield
Jana Jamsek
Suad Musovich
Markus Oscheka
Rainer Pansky
Paul Rea
Jim Sedgwick
Anthony Vandewerdt
Pete Wendler
ibm.com/redbooks
International Technical Support Organization
April 2012
SG24-7904-01
Note: Before using this information and the product it supports, read the information in Notices on
page ix.
This edition applies to Version 11 of the IBM XIV Storage System Software and Version 3 (Gen 3) of the IBM
XIV Storage System Hardware.
Copyright International Business Machines Corporation 2011, 2012. All rights reserved.
Note to U.S. Government Users Restricted Rights -- Use, duplication or disclosure restricted by GSA ADP Schedule
Contract with IBM Corp.
Contents
Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .x
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
The team who wrote this book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
Now you can become a published author, too! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv
Comments welcome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv
Stay connected to IBM Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv
Contents v
9.2.3 Assigning paths from an ESX 3.5 host to XIV. . . . . . . . . . . . . . . . . . . . . . . . . . . 271
9.3 VMware ESX and ESXi 4.x and XIV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
9.3.1 Installing HBA drivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
9.3.2 Identifying ESX host port WWN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
9.3.3 Scanning for new LUNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
9.3.4 Attaching an ESX/ESXi 4.x host to XIV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
9.3.5 Configuring ESX/ESXi 4.x host for multipathing with XIV . . . . . . . . . . . . . . . . . . 281
9.3.6 Performance tuning tips for ESX/ESXi 4.x hosts with XIV . . . . . . . . . . . . . . . . . 288
9.3.7 VMware vStorage API Array Integration (VAAI) . . . . . . . . . . . . . . . . . . . . . . . . . 292
9.4 VMware ESXi 5.0 and XIV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
9.4.1 ESXi 5.0 Fibre Channel configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
9.4.2 Performance tuning tips for ESXi 5 hosts with XIV . . . . . . . . . . . . . . . . . . . . . . . 293
9.4.3 Creating data store that are larger than 2 TiB in size . . . . . . . . . . . . . . . . . . . . . 297
9.5 VMware vStorage API Array Integration (VAAI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
9.5.1 Software prerequisites to use VAAI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
9.5.2 Installing the IBM VAAI device driver on an ESXi 4.1 server . . . . . . . . . . . . . . . 299
9.5.3 Confirming VAAI Hardware Acceleration has been detected . . . . . . . . . . . . . . . 300
9.5.4 Disabling and enabling VAAI on the XIV on a per volume basis. . . . . . . . . . . . . 304
9.5.5 Testing VAAI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
9.6 The IBM Storage Management Console for VMware vCenter . . . . . . . . . . . . . . . . . . 307
9.6.1 Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
9.6.2 Customizing the plug-in . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
9.6.3 Adding IBM Storage to the plug-in . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
9.6.4 Checking and matching XIV Volumes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
9.6.5 Creating a data store . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
9.6.6 Using a read-only user . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
9.6.7 Locating the user guide and release notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
9.6.8 Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
9.7 XIV Storage Replication Adapter for VMware SRM . . . . . . . . . . . . . . . . . . . . . . . . . . 318
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy
Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
15.1 IBM Tivoli FlashCopy Manager overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 384
15.1.1 Features of IBM Tivoli Storage FlashCopy Manager . . . . . . . . . . . . . . . . . . . . 384
15.2 Installing FlashCopy Manager 2.2.x for UNIX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
15.2.1 FlashCopy Manager prerequisites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386
15.3 Installing FlashCopy Manager with SAP in DB2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
15.3.1 FlashCopy Manager disk-only backup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
15.3.2 SAP Cloning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
15.4 Tivoli Storage FlashCopy Manager for Windows . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
15.5 Windows Server 2008 R2 Volume Shadow Copy Service . . . . . . . . . . . . . . . . . . . . 394
15.5.1 VSS architecture and components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
Contents vii
15.5.2 Microsoft Volume Shadow Copy Service function . . . . . . . . . . . . . . . . . . . . . . 397
15.6 XIV VSS Provider . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
15.6.1 Installing XIV VSS Provider . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
15.6.2 Configuring XIV VSS Provider . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399
15.6.3 Testing diskshadow VSS requester . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402
15.7 Installing Tivoli Storage FlashCopy Manager for Microsoft Exchange . . . . . . . . . . . 404
15.8 Backup scenario for Microsoft Exchange Server . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
This information was developed for products and services offered in the U.S.A.
IBM may not offer the products, services, or features discussed in this document in other countries. Consult
your local IBM representative for information about the products and services currently available in your area.
Any reference to an IBM product, program, or service is not intended to state or imply that only that IBM
product, program, or service may be used. Any functionally equivalent product, program, or service that does
not infringe any IBM intellectual property right may be used instead. However, it is the user's responsibility to
evaluate and verify the operation of any non-IBM product, program, or service.
IBM may have patents or pending patent applications covering subject matter described in this document. The
furnishing of this document does not give you any license to these patents. You can send license inquiries, in
writing, to:
IBM Director of Licensing, IBM Corporation, North Castle Drive, Armonk, NY 10504-1785 U.S.A.
The following paragraph does not apply to the United Kingdom or any other country where such
provisions are inconsistent with local law: INTERNATIONAL BUSINESS MACHINES CORPORATION
PROVIDES THIS PUBLICATION "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR
IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF NON-INFRINGEMENT,
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Some states do not allow disclaimer of
express or implied warranties in certain transactions, therefore, this statement may not apply to you.
This information could include technical inaccuracies or typographical errors. Changes are periodically made
to the information herein; these changes will be incorporated in new editions of the publication. IBM may make
improvements and/or changes in the product(s) and/or the program(s) described in this publication at any time
without notice.
Any references in this information to non-IBM websites are provided for convenience only and do not in any
manner serve as an endorsement of those websites. The materials at those websites are not part of the
materials for this IBM product and use of those websites is at your own risk.
IBM may use or distribute any of the information you supply in any way it believes appropriate without incurring
any obligation to you.
Information concerning non-IBM products was obtained from the suppliers of those products, their published
announcements or other publicly available sources. IBM has not tested those products and cannot confirm the
accuracy of performance, compatibility or any other claims related to non-IBM products. Questions on the
capabilities of non-IBM products should be addressed to the suppliers of those products.
This information contains examples of data and reports used in daily business operations. To illustrate them
as completely as possible, the examples include the names of individuals, companies, brands, and products.
All of these names are fictitious and any similarity to the names and addresses used by an actual business
enterprise is entirely coincidental.
COPYRIGHT LICENSE:
This information contains sample application programs in source language, which illustrate programming
techniques on various operating platforms. You may copy, modify, and distribute these sample programs in
any form without payment to IBM, for the purposes of developing, using, marketing or distributing application
programs conforming to the application programming interface for the operating platform for which the sample
programs are written. These examples have not been thoroughly tested under all conditions. IBM, therefore,
cannot guarantee or imply reliability, serviceability, or function of these programs.
The following terms are trademarks of the International Business Machines Corporation in the United States,
other countries, or both:
AIX Micro-Partitioning Storwize
BladeCenter Power Architecture System i
DB2 Power Systems System p
developerWorks POWER6+ System Storage
DS4000 POWER6 System x
DS8000 POWER7 System z
FICON PowerVM Tivoli
FlashCopy POWER XIV
GPFS ProtecTIER z/VM
HyperFactor pSeries z10
i5/OS Redbooks
IBM Redbooks (logo)
Intel, Intel logo, Intel Inside logo, and Intel Centrino logo are trademarks or registered trademarks of Intel
Corporation or its subsidiaries in the United States and other countries.
Ultrium, the LTO Logo and the Ultrium logo are trademarks of HP, IBM Corp. and Quantum in the U.S. and
other countries.
Microsoft, Windows, and the Windows logo are trademarks of Microsoft Corporation in the United States,
other countries, or both.
Snapshot, WAFL, Data ONTAP, and the NetApp logo are trademarks or registered trademarks of NetApp, Inc.
in the U.S. and other countries.
Java, and all Java-based trademarks and logos are trademarks or registered trademarks of Oracle and/or its
affiliates.
UNIX is a registered trademark of The Open Group in the United States and other countries.
AMD, AMD-V, the AMD Arrow logo, and combinations thereof, are trademarks of Advanced Micro Devices,
Inc.
Novell, SUSE, the Novell logo, and the N logo are registered trademarks of Novell, Inc. in the United States
and other countries.
QLogic, and the QLogic logo are registered trademarks of QLogic Corporation. SANblade is a registered
trademark in the United States.
ABAP, SAP, and SAP logos are trademarks or registered trademarks of SAP AG in Germany and in several
other countries.
Oracle, JD Edwards, PeopleSoft, Siebel, and TopLink are registered trademarks of Oracle Corporation and/or
its affiliates.
Linux is a trademark of Linus Torvalds in the United States, other countries, or both.
Other company, product, or service names may be trademarks or service marks of others.
This IBM Redbooks publication provides information for attaching the IBM XIV Storage
System to various host operating system platforms, including IBM i. The book also addresses
using the XIV storage with databases and other storage-oriented application software,
including:
IBM DB2
VMware ESX
Microsoft HyperV
SAP
The book also addresses combining the XIV Storage System with other storage platforms,
host servers, or gateways, including IBM SONAS, IBM N Series, and IBM ProtecTIER. It is
intended for administrators and architects of enterprise storage systems.
The goal is to give an overview of the versatility and compatibility of the XIV Storage System
with various platforms and environments.
The information presented here is not meant as a replacement or substitute for the Host
Attachment kit publications. It is meant as a complement and to provide readers with usage
guidance and practical illustrations. The Host Attachment kits are available at:
http://publib.boulder.ibm.com/infocenter/ibmxiv/r2/index.jsp
Bertrand Dufrasne is an IBM Certified Consulting I/T Specialist and Project Leader for IBM
System Storage disk products at the International Technical Support Organization, San
Jose Center. He has worked at IBM in various I/T areas. He has authored many IBM
Redbooks publications, and has also developed and taught technical workshops. Before
joining the ITSO, he worked for IBM Global Services as an Application Architect. He holds a
masters degree in Electrical Engineering.
Roger Eriksson is an STG Lab Services consultant, based in Stockholm, Sweden, who
works for the European Storage Competence Center in Mainz, Germany. He is a Senior
Accredited IBM Product Service Professional. Roger has over 20 years of experience working
on IBM servers and storage, including Enterprise and Midrange disk, NAS, SAN, IBM System
x, IBM System p, and IBM BladeCenter. He has done consulting, proof of concepts, and
education, mainly with the XIV product line, since December 2008. He has worked with both
clients and various IBM teams worldwide. He holds a Technical Collage Graduation in
Mechanical Engineering.
Andrew Greenfield is an IBM Field Technical Sales Engineer, based in Phoenix, Arizona,
covering the Southwest United States. He holds numerous technical certifications, notably
from Cisco, Microsoft, and IBM. He brings over 18 years inside IT at the Fortune 50 and their
data center experiences to the team. Since June 2010, Andrew has worked with consulting,
proof of concepts, and education, mainly with the IBM XIV product line. He has worked with
both clients and various IBM teams globally. He graduated Magna cum Laude, Honors
Jana Jamsek is an IT Specialist for IBM Slovenia. She works in Storage Advanced Technical
Support for Europe as an IBM Storage Systems and the IBM i (i5/OS) operating system
specialist. Jana has eight years of experience working with the IBM System i platform and
its predecessor models, and eight years of experience working with storage. She has a
masters degree in Computer Science and a degree in Mathematics from the University of
Ljubljana in Slovenia.
Suad Musovich is a XIV Technical Advisor with IBM Systems and Technology Group in New
Zealand. He has over 12 years of experience working on IBM Storage systems. He has been
involved with the planning, design, implementation, management, and problem analysis of
IBM storage solutions. His areas of expertise are the SAN infrastructure, disk storage, disk
virtualization, and IBM storage software.
Markus Oscheka is an IT Specialist for Proof of Concepts and Benchmarks in the Disk
Solution Europe team in Mainz, Germany. His areas of expertise include setup and
demonstration of IBM System Storage and TotalStorage solutions in environments such as
IBM AIX, Linux, Windows, VMware ESX, and Solaris. He has worked at IBM for nine years.
He has performed many Proof of Concepts with Copy Services on DS6000/DS8000/XIV, and
Performance-Benchmarks with DS4000/DS6000/DS8000/XIV. He has written several IBM
Redbooks and acted as the co-project lead for Redbooks including IBM DS6000/DS8000
Architecture and Implementation, DS6000/DS8000 Copy Services, and IBM XIV Storage
System: Concepts, Architecture, and Usage. He holds a degree in Electrical Engineering from
the Technical University in Darmstadt.
Paul Rea is an IBM Field Technical Sales Specialist in Littleton, Massachusetts, and covers
the North East New England Territory. He has been part of the IBM storage team focusing
primarily on the XIV product since March 2008. He has over 13 years of experience in the
storage field focusing on Engineering, Project Management, Professional Services, and
World Wide Solution Architect.
Jim Sedgwick is an IT Specialist in the US. He has more than 20 years of experience in the
storage industry. He spent five years with IBM as a printer design engineer after receiving his
Mechanical Engineering degree from NCSU. Jim's current area of expertise includes
enterprise storage performance and Copy Services. He writes and presents on both subjects.
Anthony Vandewerdt works for IBM Australia as a Storage Solutions Specialist. He has over
22 years of experience in pre- and post-sales technical support. Anthony has extensive
hands-on experience with nearly all IBM storage products, especially DS8000, Storwize
V7000, SVC, XIV, and Brocade and Cisco SANs. He has worked in a wide variety of
post-sales technical support roles including National and Asia Pacific storage support.
Anthony has also worked as an instructor and presenter for IBM STG Education.
Pete Wendler is a Software Engineer for IBM Systems and Technology Group, Storage
Platform, located in Tucson, Arizona. In his 10 years working for IBM, Peter has worked in
client support for enterprise storage products, solutions testing, and development of the IBM
DR550 archive appliance. He currently holds a position in technical marketing at IBM. Peter
received a Bachelor of Science degree from Arizona State University in 1999.
Figure 1 The 2011 XIV Gen 3 Redbooks team at the IBM ESCC Mainz, Germany
John Bynum
Iddo Jacobi
IBM
For their technical advice, support, and other contributions to this project, many thanks to:
Eyal Abraham
Dave Adams
Hanoch Ben-David
Shimon Ben-David
Alice Bird
Brian Carmody
John Cherbini
Tim Dawson
Dave Denny
Rami Elron
Orli Gan
Wilhem Gardt
Richard Heffel
Moriel Lechtman
Carlos Lizarralde
Aviad Offer
Preface xiii
Shai Priva
Joe Roa
Falk Schneider
Brian Sherman
Yossi Siles
Stephen Solewin
Anthony Vattathil
Juan Yanes
IBM
Find out more about the residency program, browse the residency index, and apply online at:
ibm.com/redbooks/residencies.html
Comments welcome
Your comments are important to us!
We want our books to be as helpful as possible. Send us your comments about this book or
other IBM Redbooks publications in one of the following ways:
Use the online Contact us review Redbooks form found at:
ibm.com/redbooks
Send your comments in an email to:
redbooks@us.ibm.com
Mail your comments to:
IBM Corporation, International Technical Support Organization
Dept. HYTD Mail Station P099
2455 South Road
Poughkeepsie, NY 12601-5400
Preface xv
xvi IBM XIV Storage System: Host Attachment and Interoperability
Summary of changes
This section describes the technical changes made in this edition of the book and in previous
editions. This edition might also include minor corrections and editorial changes that are not
identified.
Summary of Changes
for SG24-7904-01
for IBM XIV Storage System: Host Attachment and Interoperability
as created or updated on April 6, 2012.
New information
Information about the following products has been added:
VMware ESXi 5.0
Microsoft Windows 2008 R2 Cluster
Portable XIV Host Attach Kit for AIX
Changed information
The following information has changed:
SVC attachment is now covered in the IBM XIV Storage System: Copy Services and
Migration, SG24-7759.
Most chapters updates to reflect latest product information (as of November 2011).
The term host in this chapter refers to a server running a supported operating system such as
AIX or Windows. SAN Volume Controller as a host has special considerations because it acts
as both a host and a storage device. For more information, see the SAN Volume Controller
chapter in IBM XIV Storage System: Copy Services and Migration, SG24-7759.
This chapter does not address attachments from a secondary XIV used for Remote Mirroring
or data migration from an older storage system. These topics are covered in IBM XIV Storage
System: Copy Services and Migration, SG24-7759.
This chapter covers common tasks that pertain to most hosts. For operating system-specific
information regarding host attachment, see the subsequent chapters in this book.
For the latest information, see the hosts attachment kit publications at:
http://publib.boulder.ibm.com/infocenter/ibmxiv/r2/topic/com.ibm.help.xiv.doc/xiv_
pubsrelatedinfoic.html
The XIV is perfectly suited for integration into a new or existing Fibre Channel storage area
network (SAN). After the host HBAs, cabling, and SAN zoning are in place, connecting a
Fibre Channel host to an XIV is easy. The XIV storage administrator defines the hosts and
ports, and then maps volumes to them as LUNs.
You can also implement XIV with iSCSI using an existing Ethernet network infrastructure.
However, your workload might require a dedicated network. iSCSI attachment and also iSCSI
hardware initiators are not supported with all platforms. If you have Ethernet connections
between your sites, you can use that setup for a less expensive backup or disaster recovery
setup. iSCSI connections are often used for asynchronous replication to a remote site.
iSCSI-based mirroring combined with XIV snapshots or volume copies can also be used for
the following tasks:
Migrate servers between sites
Facilitate easy off-site backup or software development
The XIV Storage System has up to 15 data modules, of which up to six are also interface
modules. The number of interface modules and the activation status of the interfaces on
those modules is dependent on the rack configuration. Table 1-1 summarizes the number of
active interface modules and the FC and iSCSI ports for different rack configurations. As
shown in Table 1-1, a six module XIV physically has three interface modules, but only two of
which have active ports. An 11 module XIV physically has six interface modules, five of which
have active ports. A 2nd Generation (model A14) XIV and an XIV Gen 3 (model 114) have
different numbers of iSCSI powers.
Module 9 Not present Inactive Inactive Active Active Active Active Active
host ports ports ports
Module 8 Not present Active Active Active Active Active Active Active
host ports
Module 7 Not present Active Active Active Active Active Active Active
host ports
Fibre 8 16 16 20 20 24 24 24
Channel
Ports
iSCSI 0 4 4 6 6 6 6 6
ports on
model
A14
iSCSI 6 14 14 18 18 22 22 22
ports on
model 114
Regardless of model, each active Interface Module (Modules 4-9, if enabled) has four Fibre
Channel ports. The quantity of iSCSI ports varies based on XIV model:
For 2nd Generation XIV, up to three Interface Modules (Modules 7-9, if enabled) have two
iSCSI ports each for a maximum of six ports.
For XIV Gen 3, each active interface module except module 4 has four iSCSI ports.
Module 4 on an XIV Gen 3 has only two iSCSi ports. The maximum is therefore 22 ports.
All of these ports are used to attach hosts, remote XIVs, or older storage systems (for
migration) to the XIV. This connection can be through a SAN or iSCSI network attached to the
internal patch panel.
The patch panel simplifies cabling because the Interface Modules are pre-cabled to it.
Therefore, all your SAN and network connections are in one central place at the back of the
rack. This arrangement also helps with general cable management.
Hosts attach to the FC ports through an FC switch, and to the iSCSI ports through a Gigabit
Ethernet switch.
Restriction: Direct attachment between hosts and the XIV Storage System is not
supported.
Figure 1-1 on page 4 shows an example of how to connect to a fully populated 2nd
Generation (model A14) XIV Storage System. You can connect through either a storage area
network (SAN) or an Ethernet network. For clarity, the patch panel is not shown.
Important: Host traffic can be directed to any of the Interface Modules. The storage
administrator must ensure that host connections avoid single points of failure. The server
administrator also needs to ensure that the host workload is adequately balanced across
the connections and Interface Modules. This balancing can be done by installing the
relevant host attachment kit. Review the balancing periodically and when traffic patterns
change.
With XIV, all interface modules and all ports can be used concurrently to access any logical
volume in the system. The only affinity is the mapping of logical volumes to host, which
simplifies storage management. Balancing traffic and zoning (for adequate performance and
redundancy) is more critical, although not more complex, than with traditional storage
systems.
iSCSI
iSCSI
FCP
FCP
SAN Fabric 1 SAN Fabric 2 Ethernet Network
SI
FCP
i SC
Ethernet HBA
2 x 1 Gigabit
FC FC FC FC FC FC FC FC ETH FC FC ETH FC FC ETH ETH
Figure 1-1 Host connectivity overview with 2nd Generation XIV (without patch panel)
When connecting hosts to the XIV, there is no one size fits all solution that can be applied
because every environment is different. However, follow these guidelines avoid single points
of failure and ensure that hosts are connected to the correct ports:
FC hosts connect to the XIV patch panel FC ports 1 and 3 (or FC ports 1 and 2 depending
on your environment) on Interface Modules.
Use XIV patch panel FC ports 2 and 4 (or ports 3 and 4 depending on your environment)
for mirroring to another XIV Storage System. They can also be used for data migration
from an older storage system.
For certain environments on 2nd Generation XIV (model A14), you must use ports 1
and 2 for host connectivity and reserve ports 3 and 4 for mirroring. If you do not use
mirroring, you can also change port 4 to a target port.
Discuss with your IBM support representative what port allocation would be most
desirable in your environment.
iSCSI hosts connect to at least one port on each active Interface Module.
Restriction: A six module (27 TB) 2nd Generation XIV (model A14) does not have any
iSCSI ports. If iSCSI ports are needed, you must upgrade that XIV to a nine module
configuration or any size XIV Gen 3 (model 114).
Connect hosts to multiple separate Interface Modules to avoid a single point of failure.
Host
Module 9
SAN Fabric 1
FC FC iSCSI
Module 8
FC FC iSCSI
SAN Fabric 2
Module 7
FC FC iSCSI
Module 6 Host
Ethernet
FC FC
Network
Module 5
FC FC
Module 4
FC FC
FC
iSCSI
...191 ...193
9
...190 ...192
...180 ...182
FC FC iSCSI
...161 ...163
...171 ...173 IP(1) 6
7 ...170 ...172 IP(2) ...160 ...162
FC FC iSCSI
...151 ...153
5
...150 ...152
...161 ...163
...141 ...143
6 ...160 ...162 4 ...140 ...142
FC FC
FC FC
IP(1)
...141 ...143 8
4 ...140 ...142
IP(2)
FC FC
IP(1)
7 IP(2)
Figure 1-3 2nd Generation (model A14) patch panel to FC and iSCSI port mappings
FC (WWPN): 5001738000230xxx
IP(1)
190 192 FC iSCSI
IP(2)
9 191 193
IP(3)
190 1 1
IP(4)
FC FC iSCSI
Module 9
191 2 2
IP(1)
180 182
IP(2) 192 3 3
8 181 183
IP(3)
IP(4)
193 4 4
FC FC iSCSI
IP(1)
170 172
180 1 1
IP(2)
Interface Modules
7 IP(3)
Module 8
171 173
IP(4)
181 2 2
FC FC iSCSI
182 3 3
183 4 4
IP(1)
160 162 IP(2)
170 1 1
6 161 163
IP(3)
IP(4)
Module 7
171 2 2
FC FC iSCSI
IP(1) 172 3 3
150 152
IP(2)
5 151 153
IP(3) 173 4 4
IP(4)
FC FC iSCSI
160 1 1
IP(1)
140 142
Module 6
IP(2) 161 2 2
4 141 143
162 3 3
FC FC iSCSI
163 4 4
150 1 1
Module 5
151 2 2
152 3 3
153 4 4
140 1 1
Module 4
141 2 2
142 3
143 4
Figure 1-4 XIV Gen 3 Patch panel to FC and iSCSI port mappings
For a more detailed view of host connectivity and configuration options, see 1.2, Fibre
Channel connectivity on page 13 and 1.3, iSCSI connectivity on page 27.
To get the current list, see the IBM System Storage Interoperation Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic
From the SSIC, you can select any combination from the available boxes to determine
whether your configuration is supported. You do not have to start at the top and work down.
The result is a comma separated variable file (CSV) to show that you confirmed that your
configuration is supported.
If you cannot locate your current (or planned) combination of product versions, talk to your
IBM Business Partner, IBM Sales Representative, or IBM Pre-Sales Technical Support
Representative. You might need to request a support statement called a Storage Customer
Opportunity REquest (SCORE). It is sometimes called a request for price quotation (RPQ).
1
2
3
Host Attachment Kits are built on a Python framework and provide a consistent interface
across operating systems. Other XIV tools, such as the Microsoft Systems Operation
Manager (SCOM) management pack, also install a Python-based framework called xPYV.
With release 1.7 of the Host Attachment Kit, the Python framework is now embedded with the
Host Attachment Kit code. It is no longer a separate installer.
Before release 1.7 of the Host Attachment Kit, it was mandatory to install the Host Attachment
Kit to get technical support from IBM. Starting with release 1.7, a portable version of the Host
Attachment Kit allows all Host Attachment Kit commands to be run without installing the Host
Attachment Kit.
xiv_attach
This command locally configures the operating system and defines the host on the XIV
(except for AIX). Sometimes after running the xiv_attach command, you might be prompted
to reboot the host. This reboot can be needed because the command can perform system
modifications that force a reboot. The reboot is needed is based on the normal behavior of the
operating system. For example, a reboot is required when installing of a Windows hot fix. You
need to run this command only once, when performing initial host configuration. After the first
time, use xiv_fc_admin -R to detect newly mapped volumes.
xiv_detach
This command is used on a Windows Server to remove all XIV multipathing settings from the
host. For other operating systems, use the uninstallation option. If upgrading a server from
Windows 2003 to Windows 2008, use xiv_detach first to remove the multi-pathing settings.
The xiv_devlist command is one of the most powerful tools in your toolkit. Make sure that
you are familiar with this command and use it whenever performing system administration.
The XIV Host Attachment Kit Attachment Guide lists a number of useful parameters that can
be run with xiv_devlist. The following parameters are especially useful:
xiv_devlist -u GiB Displays the volume size in binary GB. The -u
stands for unit size.
xiv_devlist -V Displays the Host Attachment Kit version number.
The -V stands for Version.
xiv_devlist -f filename.csv -t csv Directs the output of the command to a file.
xiv_devlist -h Brings up the help page that displays other
available parameters. The -h stands for help.
xiv_diag
This command is used to satisfy requests from the IBM support center for log data. The
xiv_diag command creates a compressed packed file (using tar.gz format) that contains log
data. Therefore, you do not need to collect individual log files from your host server.
xiv_fc_admin
This command is similar to xiv_attach. Unlike xiv_attach, however, the xiv_fc_admin
command allows you to perform individual steps and tasks. The following xiv_fc_admin
command parameters are especially useful:
xiv_fc_admin -P Allows you to display the WWPNs of the host server HBAs. The -P
stands for Print.
xiv_fc_admin -V Allows you to list the tasks that xiv_attach would perform if it were
run. Knowing the tasks is vital if you are using the portable version of
the Host Attachment Kit. You need to know what tasks the Host
Attachment Kit needs to perform on your system before the change
window. The -V stands for Verify.
xiv_fc_admin -C Performs all the tasks that the xiv_fc_admin -V command identified as
being required for your operating system. The -C stands for Configure.
xiv_fc_admin -R This command scans for and configures new volumes that are
mapped to the server. For a new host that is not yet connected to an
XIV, use xiv_attach. However if additional volumes are mapped to
such a host at a later time, use xiv_fc_admin -R to detect them. You
can use native host methods but the Host Attachment Kit command is
an easier way to detect volumes. The -R stands for rescan.
xiv_fc_admin -h Brings up the help page that displays other available parameters. The
-h stands for help.
If you need co-existence and a support statement does not exist, apply for a support
statement from IBM. This statement is known as a Storage Customer Opportunity REquest
(SCORE). It is also sometimes called a request for price quotation (RPQ). There is normally
no additional charge for this support request.
Choose the connection protocol (iSCSI or FCP) based on your application requirements.
When considering IP storage-based connectivity, look at the performance and availability of
your existing infrastructure.
SAN HBA 1
FCP Fabric 1 FCP WWPN
FC
P
HBA 2
WWPN
Ethernet
iSCSI iSCSI iSCSI IQN
Network
Figure 1-6 Connecting using FCP and iSCSI simultaneously with separate host objects
A host can connect through FC and iSCSI simultaneously. However, you cannot access the
same LUN with both protocols.
Brocade
The Brocade website can be found at:
http://www.brocade.com/services-support/drivers-downloads/adapters/index.page
QLogic Corporation
The QLogic website can be found at:
http://www.qlogic.com
QLogic maintains a page that lists all the HBAs, drivers, and firmware versions that are
supported for attachment to IBM storage systems. This page can be found at:
http://support.qlogic.com/support/oem_ibm.asp
Emulex Corporation
The Emulex home page is at:
http://www.emulex.com
They also have a page with content specific to IBM storage systems at:
http://www.emulex.com/products/host-bus-adapters/ibm-branded.html
Oracle
Oracle ships its own HBAs. They are Emulex and QLogic based. However, these native
HBAs can be used to attach servers running Oracle Solaris to disk systems. In fact, such
HBAs can even be used to run StorEdge Traffic Manager software. For more information, see
the following websites:
For Emulex:
http://www.oracle.com/technetwork/server-storage/solaris/overview/emulex-corpor
ation-136533.html
For QLogic:
http://www.oracle.com/technetwork/server-storage/solaris/overview/qlogic-corp--
139073.html
HP
HP ships its own HBAs.
Emulex publishes a cross-reference at:
http://www.emulex-hp.com/interop/matrix/index.jsp?mfgId=26
QLogic publishes a cross-reference at:
http://driverdownloads.qlogic.com/QLogicDriverDownloads_UI/Product_detail.aspx?
oemid=21
This section details three typical FC configurations that are supported and offer redundancy.
All of these configurations have no single point of failure:
If a module fails, each host remains connected to all other interface modules.
If an FC switch fails, each host remains connected to at least three interface modules.
If a host HBA fails, each host remains connected to at least three interface modules.
If a host cable fails, each host remains connected to at least three interface modules.
HBA 1
HBA 2
Host 1
IBM XIV Storage System
SAN HBA 1
Fabric 1 HBA 2
Host 2
HBA 1
HBA 2
Host 3
SAN HBA 1
Host 4
Fabric 2 HBA 2
HBA 1
HBA 2
Host 5
Patch Panel
SAN Hosts
FC Ports
HBA 1
HBA 2
Host 1
IBM XIV Storage System
SAN HBA 1
Fabric 1 HBA 2
Host 2
HBA 1
HBA 2
Host 3
SAN HBA 1
Host 4
Fabric 2 HBA 2
HBA 1
HBA 2
Host 5
Patch Panel
SAN Hosts
FC Ports
HBA 1
HBA 2
Host 1
IBM XIV Storage System
SAN HBA 1
Fabric 1 HBA 2
Host 2
HBA 1
HBA 2
Host 3
SAN HBA 1
Host 4
Fabric 2 HBA 2
HBA 1
HBA 2
Host 5
Patch Panel
SAN Hosts
FC Ports
Consider the configurations shown in Table 1-2 on page 18. The columns show the interface
modules, and the rows show the number of installed modules. The table does not show how
the system is cabled to each redundant SAN fabric or how many cables are connected to the
SAN fabric. You normally connect each module to each fabric and alternate which ports you
use on each module.
For a six module system, each host has four paths per volume: Two from module 4 and
two from module 5. Port 1 on each module is connected to fabric A, whereas port 3 on
each module is connected to fabric B. Each host would be zoned to all four ports.
For a 9 or 10 module system, each host has four paths per volume: One from each
module. Port 1 on each module is connected to fabric A, whereas port 3 on each module
This path strategy works best on systems that start with nine modules. If you start with six
modules, you need to reconfigure all hosts when you upgrade to a nine module configuration.
Do not go below four paths.
1.2.3 Zoning
Zoning is mandatory when connecting FC hosts to an XIV Storage System. Zoning is
configured on the SAN switch, and isolates and restricts FC traffic to only those HBAs within
a specific zone.
A zone can be either a hard zone or a soft zone. Hard zones group HBAs depending on the
physical ports they are connected to on the SAN switches. Soft zones group HBAs depending
on the WWPNs of the HBA. Each method has its merits, and you must determine which is
right for your environment. From a switch perspective, both methods are enforced by the
hardware.
Correct zoning helps avoid issues and makes it easier to trace the cause of errors. Here are
examples of why correct zoning is important:
An error from an HBA that affects the zone or zone traffic is isolated to only the devices
that it is zoned to.
Any change in the SAN fabric triggers a registered state change notification (RSCN). Such
changes can be caused by a server restarting or a new product being added to the SAN.
An RSCN requires that any device that can see the affected or new device to
acknowledge the change, interrupting its own traffic flow.
Zoning guidelines
Zoning is affected by the following factors, among others:
Host type
Number of HBAs
HBA driver
Operating system
Applications
Therefore, it is not possible to provide a solution to cover every situation. The following
guidelines can help you to avoid reliability or performance problems. However, also review
documentation regarding your hardware and software configuration for any specific factors
that need to be considered.
Each zone (excluding those for SAN Volume Controller) has one initiator HBA (the host)
and multiple target HBA ports from a single XIV
Zone each host to ports from at least two Interface Modules.
Do not mix disk and tape traffic in a single zone. Also, avoid having disk and tape traffic on
the same HBA.
For more information about SAN zoning, see section 4.7 of Introduction to Storage Area
Networks, SG24-5470. You can download this publication from:
http://www.redbooks.ibm.com/redbooks/pdfs/sg245470.pdf
Soft zoning using the single initiator - multiple targets method is illustrated in Figure 1-10.
SAN Fabric 1
...190 ...192
...191 ...193
FCP
IBM XIV Storage System
...180 ...182
1 HBA 1 WWPN
Hosts 1
...181 ...183
2 HBA 2 WWPN
...170 ...172
...171 ...173
Spread the IO workload evenly between the interfaces. For example, for a host equipped with
two single port HBA, connect one HBA port to one port on modules 4, 6, and 8. Also, connect
When round-robin is not in use (for example, with VMware ESX 3.5 or AIX 5.3 TL9 and
earlier, or AIX 6.1 TL2 and earlier), statically balance the workload between the paths.
Monitor the IO workload on the interfaces to make sure that it stays balanced using the XIV
statistics view in the GUI (or XIVTop).
The easiest way to get a record of all the WWPNs on the XIV is to use the Extended
Command Line Interface (XCLI). However, this information is also available from the GUI.
Example 1-1 shows all WWPNs for one of the XIV Storage Systems used in the preparation
of this book. It also shows the XCLI command used to list them. For clarity, some of the
columns have been removed in this example.
Example 1-1 Getting the WWPN of an IBM XIV Storage System (XCLI)
>> fc_port_list
Component ID Status Currently WWPN Port ID Role
Functioning
1:FC_Port:4:1 OK yes 5001738000230140 00030A00 Target
1:FC_Port:4:2 OK yes 5001738000230141 00614113 Target
1:FC_Port:4:3 OK yes 5001738000230142 00750029 Target
1:FC_Port:4:4 OK yes 5001738000230143 00FFFFFF Initiator
1:FC_Port:5:1 OK yes 5001738000230150 00711000 Target
.....
1:FC_Port:6:1 OK yes 5001738000230160 00070A00 Target
....
1:FC_Port:7:1 OK yes 5001738000230170 00760000 Target
......
1:FC_Port:8:1 OK yes 5001738000230180 00060219 Target
........
1:FC_Port:9:1 OK yes 5001738000230190 00FFFFFF Target
1:FC_Port:9:2 OK yes 5001738000230191 00FFFFFF Target
1:FC_Port:9:3 OK yes 5001738000230192 00021700 Target
1:FC_Port:9:4 OK yes 5001738000230193 00021600 Initiator
The fc_port_list command might not always print out the port list in the same order.
Although they might be ordered differently, all the ports will be listed.
To get the same information from the XIV GUI, perform the following steps:
1. Select the main view of an XIV Storage System.
2. Use the arrow at the bottom (circled in red) to reveal the patch panel.
Figure 1-11 Getting the WWPNs of IBM XIV Storage System (GUI)
Tip: The WWPNs of an XIV Storage System are static. The last two digits of the
WWPN indicate to which module and port the WWPN corresponds.
As shown in Figure 1-11, the WWPN is 5001738000230160, which means that the WWPN
is from module 6, port 1. The WWPNs for the port are numbered from 0 to 3, whereas the
physical ports are numbered from 1 to 4.
The values that comprise the WWPN are shown in Example 1-2.
You typically configure 2-4 XIV ports as targets. You might need to enable the BIOS on two
HBAs, depending on the HBA, driver, and operating system. See the documentation that
came with your HBA and operating systems.
For information about SAN boot for AIX, see Chapter 4, AIX host connectivity on page 143.
For information about SAN boot for HPUX, see Chapter 5, HP-UX host connectivity on
page 177.
6. From the Configuration Settings menu, select Selectable Boot Settings to get to the
panel shown in Figure 1-16.
9. Select the IBM 2810XIV device, and press Enter to display the Select LUN menu, seen in
Figure 1-18.
11.Repeat the steps 8-10 to add additional controllers. Any additional controllers must be
zoned so that they point to the same boot LUN.
12.After all the controllers are added, press Esc to exit the Configuration Setting panel. Press
Esc again to get the Save changes option as shown in Figure 1-20.
13.Select Save changes to go back to the Fast!UTIL option panel. From there, select Exit
Fast!UTIL.
Important: Depending on your operating system and multipath drivers, you might need to
configure multiple ports as boot from SAN ports. For more information, see your
operating system documentation.
Currently, iSCSI hosts other than AIX are supported using the software iSCSI initiator. For
information about iSCSI software initiator support, see the IBM System Storage
Interoperability Center (SSIC) website at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
In the XIV Storage System, each iSCSI port is defined with its own IP address.
Redundant configurations
A redundant configuration is illustrated in Figure 1-22 on page 29.
In this configuration:
Each host is equipped with dual Ethernet interfaces. Each interface (or interface port) is
connected to one of two Ethernet switches.
Each of the Ethernet switches has a connection to a separate iSCSI port. The connection
is to Interface Modules 7-9 on a 2nd Generation XIV (model A14), and modules 4-9 on an
XIV Gen 3 (model 114).
9 Interface 2
Ethernet
7 Ethernet
Network
Interface 1
Interface 2
HOST 2
Patch Panel
Network Hosts
iSCSI Ports
Figure 1-22 iSCSI redundant configuration using 2nd Generation XIV model A14 hardware
Note: For the best performance, use a dedicated iSCSI network infrastructure.
Interface 1
9
Ethernet
IBM XIV Storage System
Network HOST 1
8
7
Interface 1
HOST 2
Patch Panel
Network Hosts
iSCSI Ports
Figure 1-23 iSCSI single network switch configuration with 2nd Generation XIV model A14 hardware
Note: Both Figure 1-22 on page 29 and Figure 1-23 show a 2nd Generation XIV (model
A14). An XIV Gen 3 has more iSCSI ports on more modules.
The IQN is visible as part of the XIV Storage System using the following steps:
1. From the XIV GUI, click Tools Settings System.
2. The Settings dialog box is displayed. Select the Parameters tab as shown in Figure 1-24.
If you are displaying multiple XIVs from the All Systems view, you can right click an XIV,
and select Properties Parameters to get the same information.
Figure 1-24 Using the XIV GUI to get the iSCSI name (IQN)
To show the same information in the XCLI, run the XCLI config_get command as shown in
Example 1-3.
Example 1-3 Using the XCLI to get the iSCSI name (IQN)
XIV PFE-Gen 3-1310133>>config_get
Name Value
dns_primary 9.64.162.21
dns_secondary 9.64.163.21
system_name XIV PFE-Gen 3-1310133
snmp_location Unknown
snmp_contact Unknown
snmp_community XIV
snmp_trap_community XIV
system_id 10133
machine_type 2810
machine_model 114
machine_serial_number 1310133
email_sender_address
email_reply_to_address
email_subject_format {severity}: {description}
iscsi_name iqn.2005-10.com.xivstorage:010133
ntp_server 9.155.70.61
support_center_port_type Management
isns_server
4. The iSCSI Connectivity window opens. Click the Define icon at the top of the window
(Figure 1-26) to open the Define IP interface dialog.
Figure 1-27 Define IP Interface - iSCSI setup window on a XIV Gen 3 (model 114)
Tip: If the MTU being used by the XIV is higher than the network can transmit, the frames
are discarded. The frames are discarded because the do-not-fragment bit is normally set
to on. Use the ping -l command to test to specify packet payload size from a Windows
workstation in the same subnet. A ping command normally contains 28 bytes of IP and
ICMP headers plus payload. Add the -f parameter to prevent packet fragmentation.
For example, the ping -f -l 1472 10.1.1.1 command sends a 1500-byte frame to the
10.1.1.1 IP address (1472 bytes of payload and 28 bytes of headers). If this command
succeeds, you can use an MTU of 1500 in the XIV GUI or XCLI.
In this example, only four iSCSI ports are configured. Non-configured ports are not displayed.
The rows might be in a different order each time you run this command. To see a complete list
of IP interfaces, use the command ipinterface_list_ports.
The CHAP configuration in the IBM XIV Storage System is defined on a per-host basis. There
are no global configurations for CHAP that affect all the hosts that are connected to the
system.
For the iSCSI initiator to log in with CHAP, both the iscsi_chap_name and iscsi_chap_secret
parameters must be set. After both of these parameters are set, the host can perform an
iSCSI login to the IBM XIV Storage System only if the login information is correct.
Configuring CHAP
Currently, you can only use the XCLI to configure CHAP. The following XCLI commands can
be used to configure CHAP:
If you are defining a new host, use the following XCLI command to add CHAP parameters:
host_define host=[hostName] iscsi_chap_name=[chapName]
iscsi_chap_secret=[chapSecret]
If the host already exists, use the following XCLI command to add CHAP parameters:
host_update host=[hostName] iscsi_chap_name=[chapName]
iscsi_chap_secret=[chapSecret]
If you no longer want to use CHAP authentication, use the following XCLI command to
clear the CHAP parameters:
host_update host=[hostName] iscsi_cha_name= iscsi_chap_secret=
9
...191 ...193 IP(1)
...181 ...183
SAN
...190 ...192 IP(2)
FC FC iSCSI ...180 ...182
Fabric 1
...181 ...183 IP(1) ...171 ...173 HBA 1
8 ...180 ...182 IP(2) ...170 ...172
WWPN: 10000000C87D295C
FC FC iSCSI HBA 2
WWPN: 10000000C87D295D
Interface Modules
...161 ...163
6 ...160 ...162
...141 ...143
...140 ...142
FC FC
The following assumptions are made for the scenario shown in Figure 1-29:
One host is set up with an FC connection. It has two HBAs and a multi-path driver
installed.
One host is set up with an iSCSI connection. It has one connection, and has the software
initiator loaded and configured.
HBA2 WWPN:
21000024FF24A427
Note: The OS Type is default for all hosts except HP-UX and IBM z/VM.
FC host-specific tasks
Configure the SAN (Fabrics 1 and 2) and power on the host server first. These actions
populate the XIV Storage System with a list of WWPNs from the host. This method is
preferable because it is less prone to error when adding the ports in subsequent procedures.
For procedures showing how to configure zoning, see your FC switch manual. The following is
an example of what the zoning details might look like for a typical server HBA zone. If you are
using SAN Volume Controller as a host, there are additional requirements that are not
addressed here.
Defining a host
To define a host, follow these steps:
1. In the XIV Storage System main GUI window, mouse over the Hosts and Clusters icon
and select Hosts and Clusters (Figure 1-30).
2. The Hosts window is displayed showing a list of hosts (if any) that are already defined. To
add a host or cluster, click either the Add Host or Add Cluster in the menu bar
(Figure 1-31). The difference between the two is that Add Host is for a single host that will
be assigned a LUN or multiple LUNs. Add Cluster is for a group of hosts that will share a
LUN or multiple LUNs.
4. To add a server to a cluster, select a cluster name. If a cluster definition was created in the
step 2, it is available in the cluster menu. In this example, no cluster was created, so None
is selected.
5. Select the Type. In this example, the type is default. If you have an HP-UX or z/VM host,
you must change the Type to match your host type. For all other hosts (such as AIX, Linux,
Solaris, VMWare, and Windows, the default option is correct).
6. Repeat steps 2 through 5 to create additional hosts if needed.
7. Host access to LUNs is granted depending on the host adapter ID. For an FC connection,
the host adapter ID is the FC HBA WWPN. For an iSCSI connection, the host adapter ID
is the host IQN. To add a WWPN or IQN to a host definition, right-click the host and select
Add Port from the menu (Figure 1-33).
Repeat steps 7 and 8 to add the second HBA WWPN. Ports can be added in any order.
9. To add an iSCSI host, specify the port type as iSCSI and enter the IQN of the HBA as the
iSCSI Name (Figure 1-35).
10.The host is displayed with its ports in the Hosts window as shown in Figure 1-36.
In this example, the hosts itso_win2008 and itso_win2008_iscsi are in fact the same
physical host. However, they are entered as separate entities so that when mapping LUNs,
the FC and iSCSI protocols do not access the same LUNs.
2. The Volume to LUN Mapping window opens as shown in Figure 1-38. Select an available
volume from the left window. The GUI suggests a LUN ID to which to map the volume.
However, this ID can be changed to meet your requirements. Click Map and the volume is
assigned immediately.
There is no difference between mapping a volume to an FC or iSCSI host in the XIV GUI
Volume to LUN Mapping view.
4. Make sure that the LUN is displayed as connected in Host Connectivity window
(Figure 1-40).
>>host_define host=itso_win2008_iscsi
Command executed successfully.
2. Host access to LUNs is granted depending on the host adapter ID. For an FC connection,
the host adapter ID is the FC HBA WWPN. For an iSCSI connection, the host adapter ID
is the IQN of the host.
In Example 1-8, the IQN of the iSCSI host is added. This is the same host_add_port
command, but with the iscsi_name parameter.
Example 1-8 Creating iSCSI port and adding it to the host definition
>> host_add_port host=itso_win2008_iscsi
iscsi_name=iqn.1991-05.com.microsoft:sand.storage.tucson.ibm.com
Command executed successfully
2. Power up the server and check the host connectivity status from the XIV Storage System
point of view. Example 1-10 shows the output for both hosts.
At this stage there might be operating system-dependent steps that need to be performed,
these steps are described in the operating system chapters.
1.5 Troubleshooting
Troubleshooting connectivity problems can be difficult. However, the XIV Storage System
does have built-in troubleshooting tools. Table 1-5 lists some of the built-in tools. For further
information, see the XCLI manual, which can be downloaded from the IBM XIV Storage
System Information Center at:
http://publib.boulder.ibm.com/infocenter/ibmxiv/r2/index.jsp
ipinterface_list_ports Lists all Ethernet ports, their configuration, and their status
Important: The procedures and instructions given here are based on code that was
available at the time of writing this book. For the latest support information and instructions,
always see the System Storage Interoperability Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
Important: The procedures and instructions given here are based on code that was
available at the time of writing this book. For the latest support information and instructions,
ALWAYS see the System Storage Interoperability Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
Also, see the XIV Storage System Host Attachment Kit Guide for Windows - Installation
Guide, which is available at:
http://publib.boulder.ibm.com/infocenter/ibmxiv/r2/topic/com.ibm.help.xiv.doc/xiv_
pubsrelatedinfoic.html
2.1.1 Prerequisites
To successfully attach a Windows host to XIV and access storage, a number of prerequisites
need to be met. The following is a generic list. Your environment might have additional
requirements.
Complete the cabling.
Complete the zoning.
Install Service Pack 1 to Windows 2008 R2.
Install hot fix KB2468345. Otherwise your server will not be able to boot again.
Create volumes to be assigned to the host.
Supported FC HBAs
Supported FC HBAs are available from Brocade, Emulex, IBM, and QLogic. Further details
about driver versions are available from the SSIC website:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
Unless otherwise noted in SSIC, use any supported driver and firmware by the HBA vendors.
For best performance, install the latest firmware and drivers for the HBAs you are using.
Multipath support
Microsoft provides a multi-path framework called Microsoft Multipath I/O (MPIO). The driver
development kit allows storage vendors to create Device Specific Modules (DSMs) for MPIO.
DSMs allow you to build interoperable multi-path solutions that integrate tightly with the
Microsoft Windows family of products. MPIO allows the host HBAs to establish multiple
sessions with the same target LUN, but present it to Windows as a single LUN. The Windows
MPIO driver enables a true active/active path policy, allowing I/O over multiple paths
With Windows operating systems, the queue depth settings are specified as part of host
adapter configuration. These settings can be specified through the BIOS settings or using
software provided by the HBA vendor.
The XIV Storage System can handle a queue depth of 1400 per FC host port, and up to 256
per volume.
Optimize your environment by evenly spreading the I/O load across all available ports. Take
into account the load on a particular server, its queue depth, and the number of volumes.
4. Follow the instructions on the panel to complete the installation. This process might
require a reboot.
5. Check that the driver is installed correctly by loading Device Manager. Verify that it now
includes Microsoft Multi-Path Bus Driver as illustrated in Figure 2-2.
The following instructions are based on the installation performed at the time of writing. For
more information, see the instructions in the Windows Host Attachment Guide. These
instructions can change over time. The instructions included here show the GUI installation.
For information about command-line instructions, see the Windows Host Attachment Guide.
Install the XIV Host Attachment Kit (it is a mandatory prerequisite for support) using these
steps:
1. Run the IBM_XIV_Host_Attachment_Kit_1.7.0-b453_for_Windows-x64.exe file. When the
setup file is run, it starts the Python engine (xpyv). Proceed with the installation following
the installation wizard instructions (Figure 2-3).
2. Run the XIV Host Attachment Wizard as shown in Figure 2-4. Click Finish to proceed.
The wizard will now validate host configuration for the XIV system.
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
Please choose a connectivity type, [f]c / [i]scsi : f
-------------------------------------------------------------------------------
Please wait while the wizard validates your existing configuration...
The wizard needs to configure the host for the XIV system.
Do you want to proceed? [default: yes ]: yes
Please wait while the host is being configured...
-------------------------------------------------------------------------------
A reboot is required in order to continue.
Please reboot the machine and restart the wizard
Example 2-2 Attaching host over FC to XIV using the XIV Host Attachment Wizard
-------------------------------------------------------------------------------
Welcome to the XIV Host Attachment wizard, version 1.7.0.
This wizard will assist you to attach this host to the XIV system.
The wizard will now validate host configuration for the XIV system.
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
Please choose a connectivity type, [f]c / [i]scsi : f
-------------------------------------------------------------------------------
Please wait while the wizard validates your existing configuration...
The wizard needs to configure the host for the XIV system.
Do you want to proceed? [default: yes ]:
Please wait while the host is being configured...
The host is now being configured for the XIV system
-------------------------------------------------------------------------------
Please zone this host and add its WWPNs with the XIV storage system:
21:00:00:24:ff:28:be:44: [QLogic QMI3572 Fibre Channel Adapter]: QMI3572
21:00:00:24:ff:28:be:45: [QLogic QMI3572 Fibre Channel Adapter]: QMI3572
21:00:00:1b:32:90:23:e7: [QLogic QMI2582 Fibre Channel Adapter]: QMI2582
21:01:00:1b:32:b0:23:e7: [QLogic QMI2582 Fibre Channel Adapter]: QMI2582
Press [ENTER] to proceed.
Would you like to rescan for new storage devices now? [default: yes ]:
Please wait while rescanning for storage devices...
-------------------------------------------------------------------------------
The XIV host attachment wizard successfully configured this host
Your Windows host now has all the required software to successfully attach to the XIV
Storage System.
The number of objects named IBM 2810XIV SCSI Disk Device depends on the number of
LUNs mapped to the host.
2. Right-click one of the IBM 2810XIV SCSI Device objects and select Properties.
The default setting here is Round Robin. Change this setting only if you are confident that
another option is better suited to your environment.
The possible options are:
Fail Over Only
Round Robin (default)
Round Robin With Subset
Least Queue Depth
Weighted Paths
IBM XIV Storage System supports the iSCSi Challenge-Handshake Authentication Protocol
(CHAP). These examples assume that CHAP is not required. If it is, specify the settings for
the required CHAP parameters on both the host and XIV sides.
To install the Windows Host Attachment Kit, use the procedure explained under Windows
Host Attachment Kit installation on page 48. Stop when you reach the step where you need
to reboot.
Example 2-3 Using the XIV Host Attachment Wizard to attach to XIV over iSCSI
Welcome to the XIV Host Attachment wizard, version 1.7.0.
This wizard will assist you to attach this host to the XIV system.
The wizard will now validate host configuration for the XIV system.
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
Please choose a connectivity type, [f]c / [i]scsi : i
-------------------------------------------------------------------------------
Please wait while the wizard validates your existing configuration...
The wizard needs to configure the host for the XIV system.
Do you want to proceed? [default: yes ]:
Please wait while the host is being configured...
-------------------------------------------------------------------------------
A reboot is required in order to continue.
Please reboot the machine and restart the wizard
Press [ENTER] to exit.
After rebooting, start the Host Attachment Kit installation wizard again, and follow the
procedure given in Example 2-4.
You can now map the XIV volumes to the defined Windows host.
2. Get the iSCSI Qualified Name (IQN) of the server from the Configuration tab (Figure 2-9).
In this example, it is iqn.1991-05.com.microsoft:win-8h202jnaffa. Copy this IQN to your
clipboard and use it to define this host on the XIV Storage System.
8. In the XIV GUI, mouse over the Hosts and Clusters icon and select iSCSI Connectivity
as shown in Figure 2-14.
The iSCSI Connectivity window shows which LAN ports are connected using iSCSI and
which IP address is used (Figure 2-15).
Alternatively, you can use the Command Line Interface (XCLI) command as shown in
Example 2-5.
The iSCSI IP addresses used in the test environment are 9.155.51.72, 9.155.51.73, and
9.155.51.74.
The FavoriteTargets tab shows the connected IP addresses as shown in Figure 2-22.
16.To see further details or change the load balancing policy, click Connections as shown in
Figure 2-24.
Figure 2-25 Windows Device Manager with XIV disks connected through iSCSI
18.Click Control Panel and select iSCSI Initiator to display the iSCSI Initiator Properties
window shown in Figure 2-27.
19.Click the Volumes and Devices tab to verify the freshly created volumes.
xiv_devlist
This utility requires Administrator privileges. The utility lists the XIV volumes available to the
host. Non-XIV volumes are also listed separately. To run it, go to a command prompt and
enter xiv_devlist as shown in Example 2-6.
xiv_diag
This utility gathers diagnostic information from the operating system. It requires Administrator
privileges. The resulting compressed file can then be sent to IBM-XIV support teams for
review and analysis. To run, go to a command prompt and enter xiv_diag as shown in
Example 2-7.
Important: The procedures and instructions given here are based on code that was
available at the time of writing. For the latest support information and instructions, see the
System Storage Interoperability Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
Also, see the XIV Storage System Host System Attachment Guide for Windows - Installation
Guide, which is available at:
http://publib.boulder.ibm.com/infocenter/ibmxiv/r2/topic/com.ibm.help.xiv.doc/xiv_
pubsrelatedinfoic.html
This section addresses the implementation of a two node Windows 2008 R2 Cluster using FC
connectivity.
If other configurations are required, you need a Storage Customer Opportunity REquest
(SCORE), which replaces the older request for price quotation (RPQ) process. IBM then tests
your configuration to determine whether it can be certified and supported. Contact your IBM
representative for more information.
Supported FC HBAs
Supported FC HBAs are available from Brocade, Emulex, IBM, and QLogic. More information
about driver versions is available from SSIC at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
Unless otherwise noted in SSIC, use any supported driver and firmware by the HBA vendors.
The latest versions are always preferred.
Multi-path support
Microsoft provides a multi-path framework and development kit called the Microsoft Multi-path
I/O (MPIO). The driver development kit allows storage vendors to create Device Specific
Modules (DSMs) for MPIO. DSMs allow you to build interoperable multi-path solutions that
integrate tightly with the Microsoft Windows.
MPIO allows the host HBAs to establish multiple sessions with the same target LUN, but
present it to Windows as a single LUN. The Windows MPIO drivers enable a true active/active
path policy allowing I/O over multiple paths simultaneously.
In this example, an XIV cluster named clu_solman has been created and both nodes are
placed in it.
2. Map all the LUNs to the cluster as shown in Figure 2-29.
All LUNs are mapped to the XIV cluster, but not to the individual nodes.
3. Set up a cluster-specific configuration that includes the following characteristics:
All nodes must be in the same domain
Have network connectivity
Private (heartbeat) network connectivity
6. Select all nodes that belong to the cluster as shown in Figure 2-32.
8. Check access to at least one of the shared drives by creating a document. For example,
create a text file on one of them, and then turn off node 1.
9. Check the access from Node2 to the shared disks and power node 1 on again.
10.Make sure that you have the right cluster witness model as illustrated in Figure 2-34. The
old cluster model had a quorum as a single point of failure.
2.2.3 Configuring the IBM Storage Enabler for Windows Failover Clustering
The IBM Storage Enabler for Windows Failover Clustering is a software agent that runs as a
Microsoft Windows Server service. It runs on two geographically dispersed cluster nodes, and
provides failover automation for XIV storage provisioning on them. This agent enables
deployment of these nodes in a geo cluster configuration.
The software, Release Notes, and the User Guide can be found at:
http://www.ibm.com/support/fixcentral/swg/selectFixes?parent=ibm~Storage_Disk&prod
uct=ibm/Storage_Disk/XIV+Storage+System+%282810,+2812%29&release=11.0.0&platform=A
ll&function=all
Table 2-1 Storage Enabler for Windows Failover Clustering supported servers
Operating system Architecture Compatibility note
Microsoft Windows Server 2003a x86, x64 Tested with R2 Service Pack 2
Microsoft Windows Server 2008a x86, x64 Tested with Service Pack 1
Microsoft Windows Server 2008 R2* x64 Tested with Service Pack 1
a. Only Enterprise and Datacenter Edition
3. Define the mirror connections for your LUNs between the two XIVs, as shown in
Figure 2-39 and for our particular configuration. For more information about how to define
the mirror pairs, see IBM XIV Storage System: Copy Services and Migration, SG 24-7759.
Figure 2-39 Mirror definitions at the master side and side of node 1
Also, define the connections on the subordinate side as shown in Figure 2-40.
Figure 2-41 Select the private mapping for node 1 on master side
Figure 2-42 shows the mapping for node 2 on the master side.
Figure 2-42 Change the default mapping: node 2 has no access to the master side
Figure 2-43 Selecting the private mapping on the subordinate side for node 2
Figure 2-44 shows the mapping for node 1 on the subordinate side.
5. Check that all resources are on node 1, where the Mirror Master side is defined, as
illustrated in Figure 2-45.
7. To deploy the resources into the geocluster, run the xiv_mcs_admin.exe utility:
C:\Program Files\XIV\mscs_agent\bin>xiv_mscs_admin.exe --deploy-resources
--verbose --xcli-username=itso --xcli-password=<PASSWORD> --yes
Figure 2-46 illustrates the cluster dependencies that result.
Figure 2-48 A switch of the cluster resources leads to a change role on XIV
A switch of the cluster resource group from node 1 to node 2 leads to a change of the
replication direction. XIV2 becomes the master and XIV1 becomes subordinate.
With the Server Core option, the setup program installs only the files needed for the
supported server roles.
Using Hyper-V on a Server Core installation reduces the attack surface. The attack surface is
the scope of interfaces, services, APIs, and protocols that a hacker can use to attempt to gain
entry into the software. As a result, a Server Core installation reduces management
requirements and maintenance. Microsoft provides management tools to remotely manage
the Hyper-V role and virtual machines (VMs). Hyper-V servers can be managed from
Windows Vista or other Windows Server 2008 or 2008 R2 systems (both x86 and x64). You
can download the management tools at:
http://support.microsoft.com/kb/952627
For more information about installation, see Implementing Microsoft Hyper-V on IBM System
x and IBM BladeCenter, REDP-4481, at:
http://www.redbooks.ibm.com/abstracts/redp4481.html?Open
Tip: The Host Attachment Kit automatically installs Microsoft Hotfix KB2460971 if
Service Pack 1 is used.
Alternatively, if no service pack is used, the following updates are automatically installed:
Microsoft Hotfix KB979711
Microsoft Hotfix KB981208
Microsoft Hotfix KB2460971
To add a role, perform the following steps. You must have local administrator privileges to add
roles to a server.
1. Click Add roles.
2. In the Before You Begin window, click Next (Figure 2-49).
4. Make sure that all the prerequisites are fulfilled before continuing as shown in Figure 2-51.
For more information about network considerations, see the Microsoft Technet website at:
http://technet.microsoft.com/en-us/library/cc816585(WS.10).aspx
After the wizard process completes (Figure 2-53), you can start using the Hyper-V role
and implement the first guests.
2. In the welcome panel, you can choose to install a default value VM guest or a custom
installation. If you will never use a default value installation, select Do not show this page
again. Click Next as shown in Figure 2-55.
4. Define the memory size of the VM, as illustrated in Figure 2-57. Click Next.
6. In the Connect Virtual Hard Disk window, define the amount of storage to provide to the
guest as C:\ and click Next (Figure 2-59).
Even if the LUN is on the XIV and you can extend the boot partition, think carefully about
the size. You need enough space for the page file, if used on C:\, and the memory dump.
8. After the wizard completes (Figure 2-61), start the VM, which boots from DVD and installs
the Windows operating system.
Figure 2-61 Guest is defined and will boot from Windows 2008 R2 SP1 DVD
2. Select the language and country-specific settings as illustrated in Figure 2-63. Click Next.
Figure 2-67 Guest is building and will boot to complete the installation
7. After the installation is finished, perform the initial configuration tasks as shown in
Figure 2-68.
This guide covers all hardware architectures that are supported for XIV attachment:
Intel x86 and x86_64, both Fibre Channel and iSCSI
IBM Power Systems
IBM System z
Older Linux versions are supported to work with the IBM XIV Storage System. However, the
scope of this chapter is limited to the most recent enterprise level distributions:
Novell SUSE Linux Enterprise Server 11, Service Pack 1 (SLES11 SP1)
Red Hat Enterprise Linux 5, Update 6 (RH-EL 5U6).
Red Hat Enterprise Linux 6, Update 1 (RH-EL 6U1).
Important: The procedures and instructions given here are based on code that was
available at the time of writing this book. For the latest support information and instructions,
ALWAYS see the System Storage Interoperability Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
All these factors make it difficult to provide generic support for Linux. As a consequence, IBM
decided on a support strategy that limits the uncertainty and the amount of testing.
IBM supports only these Linux distributions that are targeted at enterprise clients:
Red Hat Enterprise Linux (RH-EL)
SUSE Linux Enterprise Server (SLES)
These distributions have major release cycles of about 18 months, are maintained for five
years, and require you to sign a support contract with the distributor. They also have a
schedule for regular updates. These factors mitigate the issues listed previously. The limited
number of supported distributions also allows IBM to work closely with the vendors to ensure
interoperability and support. Details about the supported Linux distributions can be found in
the System Storage Interoperation Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic
IBM System z has its own web page to storage attachment using FCP at:
http://www.ibm.com/systems/z/connectivity/products/
The IBM System z Connectivity Handbook, SG24-5444, describes the connectivity options
available for use within and beyond the data center for IBM System z servers. It has a section
for FC attachment, although it is outdated with regards to multipathing. You can download this
book at:
http://www.redbooks.ibm.com/redbooks.nsf/RedbookAbstracts/sg245444.html
Past issues
The following is a partial list of storage-related issues in older Linux versions that are
addressed in recent versions:
Limited number of devices that could be attached
Gaps in LUN sequence leading to incomplete device discovery
Limited dynamic attachment of devices
Non-persistent device naming might lead to reordering
No native multipathing
In recent versions of Linux, two new subsystems were introduced, hotplug and udev. Hotplug
detects and registers newly attached devices without user intervention. Udev dynamically
creates the required device nodes for the newly attached devices according to predefined
rules. In addition, the range of major and minor numbers, the representatives of devices in the
kernel space, was increased. These numbers are now dynamically assigned.
With these improvements, the required device nodes exist immediately after a device is
detected. In addition, only device nodes that are needed are defined.
Multipathing
Linux has its own built-in multipathing solution. It is based on Device Mapper, a block device
virtualization layer in the Linux kernel. Therefore it is called Device Mapper Multipathing
(DM-MP). The Device Mapper is also used for other virtualization tasks, such as the logical
volume manager, data encryption, snapshots, and software RAID.
To remove a disk device, make sure that it is not used anymore, then remove it logically from
the system before you can physically detach it.
Write barriers are implemented in the Linux kernel using storage write cache flushes before
and after the I/O, which is order-critical. After the transaction is written, the storage cache is
flushed, the commit block is written, and the cache is flushed again. The constant flush of
caches can significantly reduce performance. You can disable write barriers at mount time
using the -o nobarrier option for mount.
Important: IBM has confirmed that Write Barriers have a negative impact on XIV
performance. Ensure that all of your mounted disks use the following switch:
mount -o nobarrier /fs
For more information, see the Red Hat Linux Enterprise 6 Storage documentation at:
http://docs.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/6/html/Storage_Administ
ration_Guide/
Direct attachment is not described because it works the same way as with the other
platforms. VIOS attachment requires specific considerations. For more information about how
VIOS works and how it is configured, see Chapter 8, IBM i and AIX clients connecting
through VIOS on page 215. Additional information is available in the following IBM
Redbooks:
IBM PowerVM Virtualization Introduction and Configuration, SG24-7940
http://www.redbooks.ibm.com/abstracts/sg247940.html
IBM PowerVM Virtualization Managing and Monitoring, SG24-7590
http://www.redbooks.ibm.com/abstracts/sg247590.html
In this example, the SCSI vendor ID is AIX, and the device model is VDASD. Apart from that,
they are treated like any other SCSI disk. If you run a redundant VIOS setup on the system,
the virtual disks can be attached through both servers. They then show up twice and must be
managed by DM-MP to ensure data integrity and path handling.
Virtual HBAs register to the SAN with their own worldwide port names (WWPNs). To the XIV
they look exactly like physical HBAs. You can create Host Connections for them and map
volumes. This process allows easier, more streamlined storage management, and better
isolation of the LPAR in an IBM Power system.
To maintain redundancy you usually use more than one virtual HBA, each one running on a
separate real HBA. Therefore, XIV volumes show up more than once (once per path) and
must be managed by a DM-MP.
System z
Linux running on an IBM System z server has the following storage attachment choices:
Linux on System z running natively in a System z LPAR
Linux on System z running in a virtual machine under z/VM
Tip: In IBM System z, the term adapters is better suited to the more common channels
term which is often used in the System z environment.
The Fibre Channel connection (IBM FICON) channel in a System z server can operate
individually in Fibre Channel Protocol (FCP) mode. FCP transports SCSI commands over the
Fibre Channel interface. It is used in all open systems implementations for SAN attached
storage. Certain operating systems that run on a System z mainframe can use this FCP
capability to connect directly to fixed block (FB) storage devices. Linux on System z provides
the kernel module zfcp to operate the FICON adapter in FCP mode. A channel can run either
in FCP or FICON mode. Channels can be shared between LPARs, and multiple ports on an
adapter can run in different modes.
To maintain redundancy you usually use more than one FCP channel to connect to the XIV
volumes. Linux sees a separate disk device for each path, and needs DM-MP to manage
them.
Tip: Linux on System z can also use count-key-data devices (CKDs). CKDs are the
traditional mainframe method to access disks. However, the XIV storage system does not
support the CKD protocol, so it is not described in this book.
Example 3-3 Loading and unloading a Linux Fibre Channel HBA Kernel module
x3650lab9:~ # modprobe qla2xxx
x3650lab9:~ # modprobe -r qla2xxx
After the driver is loaded, the FC HBA driver examines the FC fabric, detects attached
volumes, and registers them in the operating system. To discover whether a driver is loaded
or not, and what dependencies exist for it, use the command lsmod (Example 3-4).
To get detailed information about the Kernel module itself, such as the version number and
what options it supports, use the modinfo command. You can see a partial output in
Example 3-5.
Restriction: The zfcp driver for Linux on System z automatically scans and registers the
attached volumes only in the most recent Linux distributions and only if NPIV is used.
Otherwise you must tell it explicitly which volumes to access. The reason is that the Linux
virtual machine might not be intended to use all volumes that are attached to the HBA. For
more information, see Linux on System z running in a virtual machine under z/VM on
page 100, and Adding XIV volumes to a Linux on System z system on page 112.
Important:
Installing a Linux system on a SAN attached disk does not mean that it is able to start
from it. Usually you must take additional steps to configure the boot loader or boot
program.
You must take special precautions concerning multipathing if want to run Linux on SAN
attached disks.
For more information, see 3.5, Boot Linux from XIV volumes on page 137.
Linux distributions contain a script called mkinitrd that creates the initramfs image
automatically. They automatically include the HBA driver if you already used a SAN-attached
disk during installation. If not, you must include it manually. The ways to tell mkinitrd to
include the HBA driver differ depending on the Linux distribution used.
Note: The initramfs was introduced years ago and replaced the Initial RAM Disk (initrd).
Today, people often still see initrd, although they mean initramfs.
After adding the HBA driver module name to the configuration file, rebuild the initramfs with
the mkinitrd command. This command creates and installs the image file with standard
settings and to standard locations as illustrated in Example 3-7.
If you need nonstandard settings, for example a different image name, use parameters for
mkinitrd. For more information, see the man page for mkinitrd on your Linux system.
After adding the HBA driver module to the configuration file, rebuild the initramfs with the
mkinitrd command. The Red Hat version of mkinitrd requires as the following information as
parameters (Example 3-9):
The name of the image file to create
The location of the image file
The kernel version the image file is built for
If the image file with the specified name exists, use the -f option to force mkinitrd to
overwrite the existing one. The command shows more detailed output with the -v option.
You can discover the kernel version that is currently running on the system with the uname
command as illustrated in Example 3-10.
With Red Hat Enterprise Linux 6 (RH-EL6) systems, the dracut utility is always called by the
installation scripts to create an initramfs (initial RAM disk image). This process occurs
whenever a new kernel is installed using the Yum, PackageKit, or Red Hat Package Manager
(RPM).
On all architectures other than IBM i, you can create an initramfs by running the dracut
command. However, you usually do not need to create an initramfs manually. This step is
automatically performed if the kernel and its associated packages are installed or upgraded
from the RPM packages distributed by Red Hat.
Verify that an initramfs corresponding to your current kernel version exists and is specified
correctly in the grub.conf configuration file using the following procedure:
1. As root, list the contents in the /boot/ directory.
2. Find the kernel (vmlinuz-<kernel_version>) and initramfs-<kernel_version> with the most
recent version number, as shown in Figure 3-1.
Figure 3-1 Red Hat 6 (RH-EL6) display of matching initramfs and kernel
Optionally, if your initramfs-<kernel_version> file does not match the version of the latest
kernel in /boot/, or, in certain other situations, generate an initramfs file with the dracut
utility. Starting dracut as root, without options generates an initramfs file in the /boot/
directory for the latest kernel present in that directory. For more information about options and
usage, see man dracut and man dracut.conf.
On IBM i servers, the initial RAM disk and kernel files are combined into a single file created
with the addRamDisk command. This step is performed automatically if the kernel and its
associated packages are installed or upgraded from the RPM packages distributed by Red
Hat. Therefore, it does not need to be run manually.
To verify that it was created, use the command ls -l /boot/ and make sure the
/boot/vmlinitrd-<kernel_version> file exists. The <kernel_version> must match the version
of the installed kernel.
Map volumes to a Linux host as described in 1.4, Logical configuration for host connectivity
on page 35.
Tip: For Intel host systems, the XIV Host Attachment Kit can create the XIV host and host
port objects for you automatically from the Linux operating system. For more information,
see 3.2.4, Attaching XIV volumes to an Intel x86 host using the Host Attachment Kit on
page 106.
3.2.4 Attaching XIV volumes to an Intel x86 host using the Host Attachment Kit
You can attach the XIV volumes to an Intel x86 host using a Host Attachment Kit.
Attention: Although it is possible to configure Linux on Intel x86 servers manually for XIV
attachment, use the Host Attachment Kit. The Host Attachment Kit and its binary files are
required in case you need support from IBM. The kit provides data collection and
troubleshooting tools.
At the time of writing, Host Attachment Kit version 1.7 is the minimum required version for
RH-EL6.
Some additional troubleshooting checklists and tips are available in 3.4, Troubleshooting
and monitoring on page 131.
To install the Host Attachment Kit, additional Linux packages are required. These software
packages are supplied on the installation media of the supported Linux distributions. If
required software packages are missing on your host, the installation terminates. You will be
notified of the missing package.
Ensure all of the listed packages are installed on your Linux system before installing the Host
Attachment Kit.
The name of the archive, and thus the name of the directory that is created when you unpack
it, differs depending on the following items:
Your Host Attachment Kit version
Linux distribution
Hardware platform
The installation script prompts you for this information. After running the script, review the
installation log file install.log in the same directory.
The Host Attachment Kit provides the utilities you need to configure the Linux host for XIV
attachment. They are in the /opt/xiv/host_attach directory.
Remember: You must be logged in as root or have root privileges to use the Host
Attachment Kit. The Host Attachment Kit uses Python for both the installation and
uninstallation actions. Python is part of most installation distributions.
The main executables and scripts are in the directory /opt/xiv/host_attach/bin. The
installation script includes this directory in the command search path of the user root.
Therefore, the commands can be run from every working directory.
Example 3-13 Fibre Channel host attachment configuration using the xiv_attach command
[/]# xiv_attach
-------------------------------------------------------------------------------
Welcome to the XIV Host Attachment wizard, version 1.7.0
This wizard will assist you to attach this host to the XIV system.
The wizard will now validate host configuration for the XIV system.
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
iSCSI software was not detected. see the guide for more info.
Only fibre-channel is supported on this host.
Would you like to set up an FC attachment? [default: yes ]: yes
-------------------------------------------------------------------------------
Please wait while the wizard validates your existing configuration...
The wizard needs to configure the host for the XIV system.
Do you want to proceed? [default: yes ]: yes
Please wait while the host is being configured...
The host is now being configured for the XIV system
-------------------------------------------------------------------------------
Please zone this host and add its WWPNs with the XIV storage system:
10:00:00:00:c9:3f:2d:32: [EMULEX]: N/A
10:00:00:00:c9:3d:64:f5: [EMULEX]: N/A
Press [ENTER] to proceed.
Would you like to rescan for new storage devices now? [default: yes ]: yes
Please wait while rescanning for storage devices...
-------------------------------------------------------------------------------
The host is connected to the following XIV storage arrays:
Serial Ver Host Defined Ports Defined Protocol Host Name(s)
1300203 10.2 No None FC --
This host is not defined on some of the FC-attached XIV storage systems
Do you wish to define this host these systems now? [default: yes ]: yes
Configuring the host for iSCSI using the Host Attachment Kit
Use the xiv_attach command to configure the host for iSCSI attachment of XIV volumes.
First, make sure that the iSCSI service is running as illustrated in Figure 3-3.
Example 3-14 iSCSI host attachment configuration using the xiv_attach command
[/]# xiv_attach
--------------------------------------------------------------------------------
Welcome to the XIV Host Attachment wizard, version 1.7.0.
This wizard will assist you to attach this host to the XIV system.
The wizard will now validate host configuration for the XIV system.
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
Please choose a connectivity type, [f]c / [i]scsi : i
-------------------------------------------------------------------------------
Please wait while the wizard validates your existing configuration...
This host is already configured for the XIV system
-------------------------------------------------------------------------------
Would you like to discover a new iSCSI target? [default: yes ]:
Enter an XIV iSCSI discovery address (iSCSI interface): 9.155.90.183
Is this host defined in the XIV system to use CHAP? [default: no ]: no
Would you like to discover a new iSCSI target? [default: yes ]: no
Would you like to rescan for new storage devices now? [default: yes ]: yes
-------------------------------------------------------------------------------
The host is connected to the following XIV storage arrays:
Serial Ver Host Defined Ports Defined Protocol Host Name(s)
1300203 10.2 No None FC --
This host is not defined on some of the iSCSI-attached XIV storage systems.
Do you wish to define this host these systems now? [default: yes ]: yes
Please enter a name for this host [default: tic-17.mainz.de.ibm.com]: tic-17_iscsi
Please enter a username for system 1300203 : [default: admin ]: itso
Please enter the password of user itso for system 1300203:********
-------------------------------------------------------------------------------
The XIV host attachment wizard successfully configured this host
Example 3-15 shows using the Host Attachment Kit to verify the volumes for an iSCSI
attached volume. The xiv_devlist command lists all XIV devices attached to a host.
Example 3-15 Verifying mapped XIV LUNs using the Host Attachment Kit tool with iSCSI
[/]# xiv_devlist
XIV Devices
----------------------------------------------------------------------------------
Device Size Paths Vol Name Vol Id XIV Id XIV Host
Non-XIV Devices
...
Tip: The xiv_attach command already enables and configures multipathing. Therefore,
the xiv_devlist command shows only multipath devices.
If you want to see the individual devices representing each of the paths to an XIV volume, use
the lsscsi command. This command shows any XIV volumes attached to the Linux system.
Example 3-16 shows that Linux recognized 16 XIV devices. By looking at the SCSI addresses
in the first column, you can determine that there actually are four XIV volumes. Each volume
is connected through four paths. Linux creates a SCSI disk device for each of the paths.
Tip: The RH-EL installer does not install lsscsi by default. It is shipped with the
distribution, but must be selected explicitly for installation.
The nodes for XIV volumes in /dev/disk/by-id show a unique identifier. This identifier is
composed of parts of the following numbers (Example 3-17):
The worldwide node name (WWNN) of the XIV system
The XIV volume serial number in hexadecimal notation
Remember: The WWNN of the XIV system used in the examples in this chapter is
0x5001738000cb0000. It has three zeros between the vendor ID and the system ID,
whereas the representation in /dev/disk/by-id has four zeros
The udev subsystem already recognizes that there is more than one path to each XIV volume.
It creates only one node for each volume instead of four.
Important: The device nodes in /dev/disk/by-id are persistent, whereas the /dev/sdx
nodes are not. They can change when the hardware configuration changes. Do not use
/dev/sdx device nodes to mount file systems or specify system disks.
The /dev/disk/by-path file contains nodes for all paths to all XIV volumes. Here you can see
the physical connection to the volumes. This connection starts with the PCI identifier of the
HBAs, through the remote port, represented by the XIV WWPN, to the LUN of the volumes
(Example 3-18).
Attention: Due to hardware restraints, SLES10 SP3 is used for the examples. The
procedures, commands, and configuration files of other distributions can differ.
In this example, Linux on System z has two FC HBAs assigned through z/VM. Determine the
device numbers of these adapters as shown in Example 3-19.
lnxvm01:~ # lszfcp
0.0.0501 host0
0.0.0601 host1
For SLES 10, the volume configuration files are in the /etc/sysconfig/hardware directory.
There must be one for each HBA. Example 3-21 shows their naming scheme.
Attention: The configuration file described here is used with SLES9 and SLES10. SLES11
uses udev rules. These rules are automatically created by YAST when you use it to
discover and configure SAN attached volumes. They are complicated and not well
documented yet, so use YAST.
The configuration files contain a remote (XIV) port and LUN pair for each path to each
volume. Example 3-22 defines two XIV volumes to the HBA 0.0.0501, going through two XIV
host ports.
The ZFCP_LUNS=... statement in the file defines all the remote port to volume relations
(paths) that the zfcp driver sets up when it starts. The first term in each pair is the WWPN of
the XIV host port. The second term (after the colon) is the LUN of the XIV volume. The LUN
RH-EL uses the file /etc/zfcp.conf to configure SAN attached volumes. It contains the same
information in a different format, as shown in Example 3-23. The three bottom lines in the
example are comments that explain the format. They do not have to be actually present in the
file.
Today Linux has its own native multipathing solution. It is based on the Device Mapper, a
block device virtualization layer in the Linux kernel. Therefore it is called Device Mapper
Multipathing (DM-MP). The Device Mapper is also used for other virtualization tasks such as
the logical volume manager, data encryption, snapshots, and software RAID.
The /etc/multipath.conf file is not described in detail here. For more information, see the
publications in 3.1.2, Reference material on page 94. For more information about the
settings for XIV attachment, see 3.2.7, Special considerations for XIV attachment on
page 122.
One option, however, that shows up several times in the next sections needs some
explanation here. You can tell DM-MP to generate user-friendly device names by specifying
this option in /etc/multipath.conf as illustrated in Example 3-24.
The names created this way are persistent. They do not change even if the device
configuration changes. If a volume is removed, its former DM-MP name will not be used again
for a new one. If it is reattached, it gets its old name. The mappings between unique device
identifiers and DM-MP user-friendly names are stored in the file
/var/lib/multipath/bindings.
Tip: The user-friendly names are different for SLES 11 and RH-EL 5. They are explained in
their respective sections.
Important: If you install and use the Host Attachment Kit (Host Attachment Kit) on an Intel
x86 based Linux server, you do not have to set up and configure DM-MP. The Host
Attachment Kit tools configure DM-MP for you.
You can start Device Mapper Multipathing by running two start scripts as shown in
Example 3-25.
To have DM-MP start automatically at each system start, add the following start script to the
RH-EL 5 system start process (Example 3-29).
The show topology command illustrated in Example 3-30 prints out a detailed view of the
current DM-MP configuration, including the state of all available paths.
The multipath topology in Example 3-30 shows that the paths of the multipath devices are in
separate path groups. Thus, there is no load balancing between the paths. DM-MP must be
configured with a XIV multipath.conf file to enable load balancing. For more information, see
3.2.7, Special considerations for XIV attachment on page 122 and Multipathing on
page 97. The Host Attachment Kit does this configuration automatically if you use it for host
configuration.
You can use reconfigure as shown in Example 3-31 to tell DM-MP to update the topology after
scanning the paths and configuration files. Use it to add new multipath devices after adding
new XIV volumes. For more information, see section 3.3.1, Adding and removing XIV
volumes dynamically on page 123.
Attention: The multipathd -k command prompt of SLES11 SP1 supports the quit and
exit commands to terminate. That of RH-EL 5U5 is a little older and must still be
terminated using the Ctrl + d key combination.
Tip: You can also issue commands in a one-shot-mode by enclosing them in quotation
marks and typing them directly, without space, behind the multipath -k.
Although the multipath -l and multipath -ll commands can be used to print the current DM-MP
configuration, use the multipathd -k interface. The multipath tool is removed from DM-MP and
all further development and improvements go into multipathd.
This command, illustrated in Figure 3-7, enables the multipath configuration file.
Be sure to start and enable the multipathd daemon as shown in Figure 3-8.
Because the value of user_friendly_name in RH-EL6 is set to yes in the default configuration
file, the multipath devices are created as:
/dev/mapper/mpathn
Red Hat has released numerous enhancements to the device-mapper-multipath drivers that
were shipped with RH 6. Make sure to install and update to the latest version, and download
any bug fixes.
20017380000cb051f
20017380000cb0520
20017380000cb2d57
20017380000cb3af9
...
Attention: The Device Mapper itself creates its default device nodes in the /dev directory.
They are called /dev/dm-0, /dev/dm-1, and so on. These nodes are not persistent. They
can change with configuration changes and should not be used for device access.
mpath1
mpath2
mpath3
mpath4
A second set of device nodes contains the unique IDs of the volumes in their name,
regardless of whether user-friendly names are specified or not.
In RH-EL5, you find them in the directory /dev/mpath as shown in Example 3-36.
mpathd ->
../dm-7
You can also partition a DM-MP device using the fdisk command or any other partitioning
tool. To make new partitions on DM-MP devices available, use the partprobe command. It
triggers udev to set up new block device nodes for the partitions as illustrated in
Example 3-38.
Example 3-38 Using the partprobe command to register newly created partitions
x3650lab9:~ # fdisk /dev/mapper/20017380000cb051f
...
<all steps to create a partition and write the new partition table>
...
x3650lab9:~ # ls -l /dev/mapper/ | cut -c 48-
20017380000cb051f
20017380000cb0520
20017380000cb2d57
20017380000cb3af9
...
x3650lab9:~ # partprobe
x3650lab9:~ # ls -l /dev/mapper/ | cut -c 48-
20017380000cb051f
20017380000cb051f-part1
20017380000cb0520
20017380000cb2d57
20017380000cb3af9
...
Remember: This limitation, that LVM by default would not work with DM-MP devices, does
not exist in recent Linux versions.
Configuring multipathing
The XIV Host Attachment Kit normally updates the /etc/multipath.conf file during
installation to optimize use for XIV. If you need to manually update the file, the following are
the contents of this file as it is created by the Host Attachment Kit. The settings that are
relevant for XIV are shown in Example 3-39.
Important: Upgrading or reinstalling the Host Attachment Kit does not change the
multipath.conf file. Ensure that your settings match the values shown previously.
Make the changes effective by using the reconfigure command in the interactive multipathd
-k prompt.
If the version string ends with -fo, the failover capabilities are turned on and must be
disabled. To do so, add a line to the /etc/modprobe.conf file of your Linux system as
illustrated in Example 3-42.
After modifying this file, run the depmod -a command to refresh the kernel driver
dependencies. Then reload the qla2xxx module to make the change effective. If you include
the qla2xxx module in the InitRAMFS, you must create a new one.
First you discover which SCSI instances your FC HBAs have, then you issue a scan
command to their sysfs representatives. The triple dashes - - - represent the
Channel-Target-LUN combination to scan. A dash causes a scan through all possible values.
A number would limit the scan to the provided value.
Tip: If you have the Host Attachment Kit installed, you can use the xiv_fc_admin -R
command to scan for new XIV volumes.
New disk devices that are discovered this way automatically get device nodes and are added
to DM-MP.
Tip: For some older Linux versions, you must force the FC HBA perform a port login to
recognize newly added devices. Use the following command, which must be issued to all
FC HBAs:
echo 1 > /sys/class/fc_host/host<ID>/issue_lip
If you want to remove a disk device from Linux, follow this sequence to avoid system hangs
due to incomplete I/O requests:
1. Stop all applications that use the device and make sure that all updates or writes are
completed.
2. Unmount the file systems that use the device.
3. If the device is part of an LVM configuration, remove it from all logical volumes and volume
groups.
4. Remove all paths to the device from the system (Example 3-44).
The device paths (or disk devices) are represented by their Linux SCSI address. FOr more
information, see Linux SCSI addressing explained on page 110. Run the multipathd
-kshow topology command after removing each path to monitor the progress.
DM-MP and udev recognize the removal automatically and delete all corresponding disk and
multipath device nodes. You must remove all paths that exist to the device before detaching
the device on the storage system level.
You can use watch to run a command periodically for monitoring purposes. This example
allows you to monitor the multipath topology with a period of one second:
Example 3-45 shows how to discover and list the connected volumes, in this case one remote
port or path, with the zfcp_san_disc command. You must run this command for all available
remote ports.
After discovering the connected volumes, logically attach them using sysfs interfaces.
Remote ports or device paths are represented in the sysfs. There is a directory for each local
- remote port combination (path). It contains a representative of each attached volume and
various meta files as interfaces for action. Example 3-46 shows such a sysfs structure for a
specific XIV port.
Add LUN 0x0003000000000000 to both available paths using the unit_add metafile as shown
in Example 3-47.
Attention: You must perform discovery using zfcp_san_disc whenever new devices,
remote ports, or volumes are attached. Otherwise the system does not recognize them
even if you do the logical configuration.
New disk devices that you attached this way automatically get device nodes and are added to
DM-MP.
If you want to remove a volume from Linux on System z, perform the same steps as for the
other platforms. These procedures avoid system hangs due to incomplete I/O requests:
Volumes can then be removed logically, using a method similar to the attachment. Write the
LUN of the volume into the unit_remove meta file for each remote port in sysfs.
Important: If you need the newly added devices to be persistent, use the methods shown
in Adding XIV volumes to a Linux on System z system on page 112. Create the
configuration files to be used at the next system start.
Attach the new XIV ports logically to the HBAs. In Example 3-49, a remote port is already
attached to HBA 0.0.0501. Add the second connected XIV port to the HBA.
Add the second new port to the other HBA in the same way (Example 3-50).
Example 3-51 Checking the size and available space on a mounted file system
x3650lab9:~ # df -h /mnt/itso_0520/
file system Size Used Avail Use% Mounted on
/dev/mapper/20017380000cb0520
16G 173M 15G 2% /mnt/itso_0520
2. Use the XIV GUI to increase the capacity of the volume from 17 to 51 GB (decimal, as
shown by the XIV GUI). The Linux SCSI layer picks up the new capacity when you rescan
each SCSI disk device (path) through sysfs (Example 3-52).
The message log shown in Example 3-53 indicates the change in capacity.
Example 3-53 Linux message log indicating the capacity change of a SCSI device
x3650lab9:~ # tail /var/log/messages
...
Oct 13 16:52:25 lnxvm01 kernel: [ 9927.105262] sd 0:0:0:4: [sdh] 100663296
512-byte logical blocks: (51.54 GB/48 GiB)
Oct 13 16:52:25 lnxvm01 kernel: [ 9927.105902] sdh: detected capacity change
from 17179869184 to 51539607552
...
3. Indicate the device change to DM-MP using the resize_map command of multipathd. The
updated capacity is displayed in the output of show topology (Example 3-54).
4. Resize the file system and check the new capacity as shown in Example 3-55.
x3650lab9:~ # df -h /mnt/itso_0520/
file system Size Used Avail Use% Mounted on
/dev/mapper/20017380000cb0520
48G 181M 46G 1% /mnt/itso_0520
Restriction: At the time of writing, the dynamic volume increase process has the following
restrictions:
Of the supported Linux distributions, only SLES11 SP1 has this capability. The
upcoming RH-EL 6 will also have it.
The sequence works only with unpartitioned volumes.
The file system must be created directly on the DM-MP device.
Only the modern file systems can be resized while they are mounted. The ext2 file
system cannot.
This section describes some methods to avoid integrity issues. It also highlights some
potential traps that might lead to problems.
2. Make sure that the data on the source volume is consistent by running the sync command.
3. Create the snapshot on the XIV, make it writeable, and map the target volume to the Linux
host. In the example, the snapshot source has the volume ID 0x0520, and the target
volume has ID 0x1f93.
5. Mount the target volume to a separate mount point using a device node created from the
unique identifier of the volume (Example 3-58).
Now you can access both the original volume and the point-in-time copy through their
respective mount points.
Attention: udev also creates device nodes that relate to the file system unique identifier
(UUID) or label. These IDs are stored in the data area of the volume, and are identical on
both source and target. Such device nodes are ambiguous if the source and target are
mapped to the host at the same time. Using them in this situation can result in data loss.
A script called vgimportclone.sh is publicly available that automates the modification of the
metadata. It can be downloaded from:
http://sources.redhat.com/cgi-bin/cvsweb.cgi/LVM2/scripts/vgimportclone.sh?cvsroot
=lvm2
An online copy of the Linux man page for the script can be found at:
http://www.cl.cam.ac.uk/cgi-bin/manpage?8+vgimportclone
Tip: The vgimportclone script and commands are part of the standard LVM tools for
RH-EL. The SLES 11 distribution does not contain the script by default.
2. Make sure that the data on the source volume is consistent by running the sync command.
3. Create the snapshots on the XIV, unlock them, and map the target volumes 1fe4 and 1fe5
to the Linux host.
4. Initiate a device scan on the Linux host. For more information, see 3.3.1, Adding and
removing XIV volumes dynamically on page 123. DM-MP automatically integrates the
snapshot targets as shown in Example 3-60.
5. Run the vgimportclone.sh script against the target volumes, and providing a new volume
group name (Example 3-61).
6. Activate the volume group on the target devices and mount the logical volume as shown in
Example 3-62.
Example 3-62 Activating volume group on target device and mount the logical volume
x3650lab9:~ # vgchange -a y vg_itso_snap
1 logical volume(s) in volume group "vg_itso_snap" now active
x3650lab9:~ # mount /dev/vg_itso_snap/lv_itso /mnt/lv_snap_itso/
x3650lab9:~ # mount
...
/dev/mapper/vg_xiv-lv_itso on /mnt/lv_itso type ext3 (rw)
/dev/mapper/vg_itso_snap-lv_itso on /mnt/lv_snap_itso type ext3 (rw)
Afterward, key information can be found in the same directory the installation was started
from, inside the install.log file.
Example 3-63 Options of xiv_devlist from Host Attachment Kit version 1.7
# xiv_devlist --help
Usage: xiv_devlist [options]
-h, --help show this help message and exit
-t OUT, --out=OUT Choose output method: tui, csv, xml (default: tui)
-o FIELDS, --options=FIELDS
Fields to display; Comma-separated, no spaces. Use -l
to see the list of fields
-f OUTFILE, --file=OUTFILE
File to output to (instead of STDOUT) - can be used
only with -t csv/xml
-H, --hex Display XIV volume and machine IDs in hexadecimal base
-u SIZE_UNIT, --size-unit=SIZE_UNIT
The size unit to use (e.g. MB, GB, TB, MiB, GiB, TiB,
...)
-d, --debug Enable debug logging
-l, --list-fields List available fields for the -o option
xiv_diag
This utility gathers diagnostic information from the operating system. The resulting
compressed file can then be sent to IBM-XIV support teams for review and analysis
(Example 3-64).
For more detailed information, use the multipath -v2 -d as illustrated in Example 3-65
Another excellent command-line utility to use is the xiv_devlist command. Note that
Example 3-66 shows paths that were not found in Figure 3-9 on page 134, and vice versa.
XIV Devices
-------------------------------------------------------------------------------
Device Size (GB) Paths Vol Name Vol Id XIV Id XIV Host
-------------------------------------------------------------------------------
/dev/sdc 51.6 N/A RedHat-Data_1 593 1310133 RedHat6.de.ibm.com
-------------------------------------------------------------------------------
/dev/sdd 51.6 N/A RedHat-Data_2 594 1310133 RedHat6.de.ibm.com
-------------------------------------------------------------------------------
/dev/sde 51.6 N/A RedHat-Data_1 593 1310133 RedHat6.de.ibm.com
-------------------------------------------------------------------------------
/dev/sdf 51.6 N/A RedHat-Data_2 594 1310133 RedHat6.de.ibm.com
-------------------------------------------------------------------------------
Non-XIV Devices
--------------------------
Device Size (GB) Paths
--------------------------
/dev/sda 50.0 N/A
--------------------------
/dev/sdb 50.0 N/A
--------------------------
Figure 3-9 Second example of xiv_devlist command, showing multipath not working properly
Important: When using the xiv_devlist command, note the number of paths indicated in
the column for each device. You do not want the xiv_devlist output to show N/A in the
paths column.
The expected output from the multipath and xiv_devlist commands is shown in
Example 3-67.
Example 3-67 Example of multipath finding the XIV devices, and updating the paths correctly
[root@bc-h-15-b7 ~]#multipath
create: mpathc (20017380027950251) undef IBM,2810XIV
size=48G features='0' hwhandler='0' wp=undef
`-+- policy='round-robin 0' prio=1 status=undef
|- 8:0:0:1 sdc 8:32 undef ready running
`- 9:0:0:1 sde 8:64 undef ready running
create: mpathd (20017380027950252) undef IBM,2810XIV
size=48G features='0' hwhandler='0' wp=undef
`-+- policy='round-robin 0' prio=1 status=undef
|- 8:0:0:2 sdd 8:48 undef ready running
`- 9:0:0:2 sdf 8:80 undef ready running
XIV Devices
-------------------------------------------------------------------------------
Device Size (GB) Paths Vol Name Vol Id XIV Id XIV Host
Non-XIV Devices
--------------------------
Device Size (GB) Paths
--------------------------
/dev/sda 50.0 N/A
--------------------------
/dev/sdb 50.0 N/A
--------------------------
...
Remember: For Linux on System z under z/VM, the OS loader is not part of the
firmware. Instead, it is part of the z/VM program ipl.
A detailed description of the Linux boot process for x86 based systems can be found on IBM
developerWorks at:
http://www.ibm.com/developerworks/linux/library/l-linuxboot/
Tip: Emulex HBAs also support booting from SAN disk devices. You can enable and
configure the Emulex BIOS extension by pressing Alt+e or Ctrl+e when the HBAs are
initialized during server startup. For more information, see the following Emulex
publications:
Supercharge Booting Servers Directly from a Storage Area Network
http://www.emulex.com/artifacts/fc0b92e5-4e75-4f03-9f0b-763811f47823/booting
ServersDirectly.pdf
Enabling Emulex Boot from SAN on IBM BladeCenter
http://www.emulex.com/artifacts/4f6391dc-32bd-43ae-bcf0-1f51cc863145/enablin
g_boot_ibm.pdf
IBM System z
Linux on System z can be loaded from traditional CKD disk devices or from Fibre-Channel-
attached Fixed-Block (SCSI) devices. To load from SCSI disks, the SCSI IPL feature (FC
9904) must be installed and activated on the System z server. The SCSI Initial Program
Load (IPL) is generally available on recent System z systems (IBM z10 and later)
CP QUERY LOADDEV
PORTNAME 50017380 00CB0191 LUN 00010000 00000000 BOOTPROG 0
BR_LBA 00000000 00000000
The port name is the XIV host port used to access the boot volume. After the load device
is set, use the IPL program with the device number of the FCP device (HBA) that connects
to the XIV port and LUN to boot from. You can automate the IPL by adding the required
commands to the z/VM profile of the virtual machine.
Tip: After the SLES11 installation program (YAST) is running, the installation is mostly
hardware platform independent. It works the same when running on an x86, IBM Power
System, or System z server.
Remember: The Linux on System z installer does not automatically list the available
disks for installation. Use the Configure Disks window to discover and attach the disks
that are needed to install the system using a graphical user interface. This window is
displayed before you get to the Installation Settings window. At least one disk device
is required to perform the installation.
2. Click Partitioning.
3. In the Preparing Hard Disk: Step 1 window, make sure that Custom Partitioning (for
experts) is selected and click Next (Figure 3-11). It does not matter which disk device is
selected in the Hard Disk field.
5. Confirm your selecting, and the tool rescans the disk devices. When finished, it presents
an updated list of hard disks that also shows the multipath devices it found (Figure 3-13).
6. Select the multipath device (XIV volume) you want to install to and click Accept.
7. In the Partitioner window, create and configure the required partitions for your system the
same way you would on a local disk.
You can also use the automatic partitioning capabilities of YAST after the multipath devices
are detected in step 5. To do so, perform the following steps:
1. Click Back until you see the initial partitioning window again. It now shows the multipath
devices instead of the disks, as illustrated in Figure 3-14.
Figure 3-14 Preparing Hard Disk: Step 1 window with multipath devices
2. Select the multipath device you want to install on and click Next.
3. Select the partitioning scheme you want.
The installer does not implement any device-specific settings, such as creating the
/etc/multipath.conf file. You must do implement these settings manually after installation as
explained in 3.2.7, Special considerations for XIV attachment on page 122. Because
DM-MP is already started during the processing of the InitRAMFS, you also must build a new
InitRAMFS image after changing the DM-MP configuration. For more information, see Making
the FC driver available early in the boot process on page 103.
It is possible to add Device Mapper layers on top of DM-MP, such as software RAID or LVM.
The Linux installers support these options.
Tip: RH-EL 5.1 and later support multipathing already. Turn on multipathing it by adding
the option mpath to the kernel boot line of the installation system. Anaconda, the RH
installer, then offers to install to multipath devices
Important: The procedures and instructions given here are based on code that was
available at the time of writing. For the latest support information and instructions, see the
System Storage Interoperability Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
The AIX host attachment process with XIV is described in detail in the Host Attachment Guide
for AIX which is available from the XIV information center at:
http://publib.boulder.ibm.com/infocenter/ibmxiv/r2/topic/com.ibm.help.xiv.doc/xiv_
pubsrelatedinfoic.html
The XIV Storage System supports different versions of the AIX operating system through
either Fibre Channel (FC) or iSCSI connectivity.
Important: The procedures and instructions given here are based on code that was
available at the time of writing this book. For the latest support information and instructions,
see the System Storage Interoperability Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
4.1.1 Prerequisites
If the current AIX operating system level installed on your system is not compatible with XIV,
you must upgrade before attaching the XIV storage. To determine the maintenance package
or technology level currently installed on your system, use the oslevel command
(Example 4-1).
In this example, the system is running AIX 6.1.0.0 technology level 5 (61TL5). Use this
information in conjunction with the SSIC to ensure that you have an IBM-supported
configuration.
If AIX maintenance items are needed, consult the IBM Fix Central website. You can download
fixes and updates for your systems software, hardware, and operating system at:
http://www.ibm.com/eserver/support/fixes/fixcentral/main/pseries/aix
Before further configuring your host system or the XIV Storage System, make sure that the
physical connectivity between the XIV and the POWER system is properly established. Direct
attachment of XIV to the host system is not supported. If using FC switched connections,
ensure that you have functioning zoning using the worldwide port name (WWPN) numbers of
the AIX host.
The lsslot command returns not just the ports, but also the PCI slot where the Fibre Channel
adapters are in the system (Example 4-3). This command can be used to identify in what
physical slot a specific adapter is placed.
To obtain the WWPN of each of the POWER system FC adapters, use the lscfg commands
as shown in Example 4-4.
Part Number.................10N9824
Serial Number...............1B839041E7
Manufacturer................001B
EC Level....................D76482A
Customer Card ID Number.....577D
FRU Number..................10N9824
Device Specific.(ZM)........3
Network Address.............10000000C9808610
ROS Level and ID............02781115
Device Specific.(Z0)........31004549
Device Specific.(Z1)........00000000
Device Specific.(Z2)........00000000
Device Specific.(Z3)........09030909
Device Specific.(Z4)........FF781110
You can also print the WWPN of an HBA directly by issuing this command:
lscfg -vl <fcs#> | grep Network
You can now define the AIX host system on the XIV Storage System and assign the WWPN
to the host. These ports are selectable from the list as shown in Figure 4-1 if the following
conditions are true:
The FC connection is correctly done
The zoning is enabled
The Fibre Channel adapters are in an available state on the host
After creating the AIX host, map the XIV volumes to the host.
Figure 4-1 Selecting ports from the Port Name menu in the XIV GUI
Tip: If the WWPNs are not displayed in the list, run the cfgmgr command on the AIX host to
activate the HBAs. If you still do not see the WWPNs, remove the fcsX with the command
rmdev -Rdl fcsX, then run cfgmgr again.
With older AIX releases, you might see a warning when you run the cfgmgr or xiv_fc-admin
-R command as shown in Example 4-5. This warning can be ignored. You can also find a fix
for this erroneous warning at:
http://www.ibm.com/support/docview.wss?uid=isg1IZ75967
Important: Although AIX now natively supports XIV using ODM changes that have been
back-ported to several older AIX releases, install the XIV Host Attachment Kit. The kit
provides support and access to the latest XIV utilities like xiv_diag. The output of these
XIV utilities is mandatory for IBM support when opening an XIV-related service call.
Would you like to proceed and install the Host Attachment Kit? [Y/n]:
y
Please wait while the installer validates your existing configuration...
---------------------------------------------------------------
Please wait, the Host Attachment Kit is being installed...
---------------------------------------------------------------
Installation successful.
Please see the Host Attachment Guide for information about how to configure
this host.
When the installation completes, listing the disks should display the correct number of disks
seen from the XIV storage. They are labeled as XIV disks as illustrated in Example 4-7.
The wizard will now validate host configuration for the XIV system.
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
Please choose a connectivity type, [f]c / [i]scsi : fc
-------------------------------------------------------------------------------
Please wait while the wizard validates your existing configuration...
This host is already configured for the XIV system
-------------------------------------------------------------------------------
Please zone this host and add its WWPNs with the XIV storage system:
10:00:00:00:C9:80:86:10: fcs0: [IBM]: N/A
10:00:00:00:C9:80:86:11: fcs1: [IBM]: N/A
Press [ENTER] to proceed.
Would you like to rescan for new storage devices now? [default: yes ]: y
Please wait while rescanning for storage devices...
-------------------------------------------------------------------------------
No XIV LUN0 devices were detected
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
The XIV host attachment wizard successfully configured this host
Important: The No XIV LUN0 devices were detected message is displayed by default.
This message can be ignored. This section is intended for when AIX supports in-band
management. It is not supported at this time.
LUNs that are seen by the AIX server can now be used. Do not run the xiv_attach command
more than once. If more LUNS are added in the future, then use the xiv_fc_admin -R to scan
for the new LUNs.
Example 4-9 shows the output of lsdev after the XIV volume was removed in the XIV GUI,
which was visible as hdisk1 on the AIX server. Notice how the LUN is still displayed in the list.
# xiv_devlist
XIV Devices
-------------------------------------------------------------------------------
Device Size (GB) Paths Vol Name Vol Id XIV Id XIV Host
-------------------------------------------------------------------------------
/dev/hdisk2 1032.5 2/2 CUS_Lisa_143 232 1310114 AIX_P570_2_lpar2
-------------------------------------------------------------------------------
/dev/hdisk3 1032.5 2/2 CUS_Zach 231 1310114 AIX_P570_2_lpar2
-------------------------------------------------------------------------------
Non-XIV Devices
-----------------------------
Device Size (GB) Paths
-----------------------------
/dev/hdisk0 32.2 1/1
-----------------------------
Unreachable devices: /dev/hdisk1
In some instances, the devices might not be removed. If that happens, use the rmdev
command to remove the devices. The command and syntax is shown in Example 4-12. To
determine which hdisk needs to be removed, run lsdev -Cc disk to show the hdisks
(Example 4-11).
Example 4-11 lsdev command output after the devices are removed
# lsdev -Cc disk
hdisk0 Available Virtual SCSI Disk Drive
hdisk1 Available 00-01-02 MPIO 2810 XIV Disk
hdisk2 Available 00-01-02 MPIO 2810 XIV Disk
hdisk3 Available 00-01-02 MPIO 2810 XIV Disk
If the disks were unmapped but the lsdev command still shows them, use the rmdev
command with the d and l flags (Example 4-12). The rmdev command unconfigures and
undefines the device specified with the device logical name using the -l Name flag. The -d
flag deletes the device definition from the Customized Devices object class. Do not run this
command for devices that are in production.
After running the rmdev command, use the lsdev and xiv_devlist commands to verify that
the devices are removed.
Other AIX commands such as cfgmgr can also be used, but these commands are run within
the XIV commands.
The xiv_fc_admin command can be used to confirm that the AIX server is running a
supported configuration and ready to attach to the XIV storage. Use the xiv_fc_admin -V
command to verify the configuration and be notified if any OS component is missing. The
xiv_attach command must be run the first time the server is attached to the XIV array. It is
used to scan for new XIV LUNs and configure the server to work with XIV. Do not run the
xiv_attach command more than once. If more LUNS are added in the future, use the
xiv_fc_admin -R to scan for the new LUNs. All of these commands and the others in the
portable Host Attachment Kit are defined in 4.1.5, Host Attachment Kit utilities on page 166.
To use the portable Host Attachment Kit package from a network drive:
1. Extract the files from <HAK_build_name>_Portable.tar.gz into a shared folder on a
network drive.
2. Mount the shared folder to each host computer you intend to use the Host Attachment Kit
on. The folder must be recognized and accessible as a network drive.
You can now use the IBM XIV Host Attachment Kit on any host to which the network drive is
mounted.
To run commands from the portable Host Attachment Kit location, use ./ before every
command. Example 4-13 shows the xiv_attach command and the output when run from the
portable Host Attachment Kit location.
The wizard will now validate host configuration for the XIV system.
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
Please choose a connectivity type, [f]c / [i]scsi : fc
-------------------------------------------------------------------------------
Please wait while the wizard validates your existing configuration...
This host is already configured for the XIV system
-------------------------------------------------------------------------------
Please zone this host and add its WWPNs with the XIV storage system:
10:00:00:00:C9:80:86:10: fcs0: [IBM]: N/A
10:00:00:00:C9:80:86:11: fcs1: [IBM]: N/A
Would you like to rescan for new storage devices now? [default: yes ]: y
Please wait while rescanning for storage devices...
-------------------------------------------------------------------------------
No XIV LUN0 devices were detected
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
The XIV host attachment wizard successfully configured this host
Tip: Whenever a newer Host Attachment Kit version is installed on the network drive, all
hosts to which that network drive was mounted have access to that version
Important: For more information about setting up servers that use the portable Host
Attachment Kit, see AIX MPIO on page 153.
If the Host Attachment Kit is locally installed on the host, you can uninstall it without detaching
the host from XIV.
The portable Host Attachment Kit packages do not require the uninstallation procedure. You
can delete the portable Host Attachment Kit directory on the network drive or the USB flash
drive to uninstall it. For more information about the portable Host Attachment Kit, see the
previous section.
The regular uninstallation removes the locally installed Host Attachment Kit software without
detaching the host. This process preserves all multipathing connections to the XIV storage
system.
Use the following command to uninstall the Host Attachment Kit software:
# /opt/xiv/host_attach/bin/uninstall
If you get the message Please use the O/S package management services to remove the
package, use the package management service to remove the Host Attachment Kit. The
SUCCESSES
---------
file sets listed in this section passed pre-deinstall verification
and will be removed.
+-----------------------------------------------------------------------------+
Deinstalling Software...
+-----------------------------------------------------------------------------+
+-----------------------------------------------------------------------------+
Summaries:
+-----------------------------------------------------------------------------+
Installation Summary
--------------------
Name Level Part Event Result
-------------------------------------------------------------------------------
xiv.hostattachment.tools 1.7.0.0 USR DEINSTALL SUCCESS
The MPIO base functionality is limited. It provides an interface for vendor-specific path control
modules (PCMs) that allow for implementation of advanced algorithms.
For more information, see the IBM pSeries and AIX Information Center at:
http://publib.boulder.ibm.com/infocenter/aix/v7r1/index.jsp
AIX provides the manage_disk_drivers command to switch a device between MPIO and
non-MPIO. This command can be used to change how the XIV device is configured. All XIV
disks are converted.
Important: It is not possible to convert one XIV disk to MPIO and another XIV disk to
non-MPIO.
To switch XIV 2810 devices from MPIO to non-MPIO, run the following command:
manage_disk_drivers -o AIX_non_MPIO -d 2810XIV
To switch XIV 2810 devices from non-MPIO to MPIO, run the following command:
manage_disk_drivers -o AIX_AAPCM -d 2810XIV
After running either of these commands, the system will need to be rebooted for the
configuration change to take effect.
If the application is I/O intensive and uses large block I/O, the queue_depth and the max
transfer size might need to be adjusted. Such an environment typically needs a queue_depth
of 64 - 256, and max_tranfer=0x100000. Typical values are 40 - 64 as the queue depth per
LUN, and 512-2048 per HBA in AIX.
The changes can be confirmed by running the lsattr command again (Example 4-18).
To check the queue depth, periodically run iostat -D 5. If avgwqsz (average wait queue size)
or sqfull are consistently greater zero, increase the queue depth. The maximum queue
depth is 256. However, do not start at 256 and work down because you might flood the XIV
with commands. 64 is a good number for the vast majority of environments.
See the following tables for the minimum level of service packs and the Host Attachment Kit
version needed for each AIX version.
Table 4-1 shows the service packs and Host Attachment Kit version needed for AIX 5.3.
Table 4-1 AIX 5.3 minimum level service packs and Host Attachment Kit versions
AIX Release Technology Level Service pack Host Attachment Kit
Version
Table 4-2 AIX 6.1 minimum level service packs and Host Attachment Kit versions
AIX Release Technology Level Service pack Host Attachment Kit
Version
Table 4-3 shows the service packs and Host Attachment Kit version needed for AIX 7.1.
Table 4-3 AIX 7.1 minimum level service pack and Host Attachment Kit version
AIX Release Technology Level Service pack Host Attachment Kit
Version
The default disk behavior algorithm is round_robin with a queue depth of 40. If the
appropriate AIX technology level and service packs list are met, this queue depth restriction is
lifted and the settings can be adjusted.
Example 4-19 shows how to adjust the disk behavior algorithm and queue depth setting.
Example 4-19 Changing disk behavior algorithm and queue depth command
# chdev -a algorithm=round_robin -a queue_depth=40 -l <hdisk#>
If you want the fail_over disk behavior algorithm, load balance the I/O across the FC adapters
and paths. Set the path priority attribute for each LUN so that 1/nth of the LUNs are assigned
to each of the n FC paths.
To change the path priority of an MPIO device, use the chpath command. An example of this
process is shown in Example 4-22 on page 157.
Initially, use the lspath command to display the operational status for the paths to the devices
as shown in Example 4-20.
Example 4-20 The lspath command shows the paths for hdisk2
# lspath -l hdisk2 -F status:name:parent:path_id:connection
Enabled:hdisk2:fscsi0:0:5001738000130140,2000000000000
Enabled:hdisk2:fscsi0:1:5001738000130150,2000000000000
Enabled:hdisk2:fscsi0:2:5001738000130160,2000000000000
Enabled:hdisk2:fscsi0:3:5001738000130170,2000000000000
Enabled:hdisk2:fscsi0:4:5001738000130180,2000000000000
Enabled:hdisk2:fscsi0:5:5001738000130190,2000000000000
Enabled:hdisk2:fscsi1:6:5001738000130142,2000000000000
Enabled:hdisk2:fscsi1:7:5001738000130152,2000000000000
Enabled:hdisk2:fscsi1:8:5001738000130162,2000000000000
Enabled:hdisk2:fscsi1:9:5001738000130172,2000000000000
Enabled:hdisk2:fscsi1:10:5001738000130182,2000000000000
Enabled:hdisk2:fscsi1:11:5001738000130192,2000000000000
The lspath command can also be used to read the attributes of a path to an MPIO capable
device as shown in Example 4-21. The <connection> info is either <SCSI ID>, <LUN ID> for
SCSI (for example 5, 0) or <WWN>, <LUN ID> for FC devices (Example 4-21).
Example 4-21 The lspath command reads attributes of the 0 path for hdisk2
# lspath -AHE -l hdisk2 -p fscsi0 -w "5001738000130140,2000000000000"
attribute value description user_settable
As noted, the chpath command is used to perform change operations on a specific path. It
can either change the operational status or tunable attributes associated with a path. It cannot
perform both types of operations in a single invocation.
Example 4-22 illustrates the use of the chpath command with an XIV Storage System. The
command sets the primary path to fscsi0 using the first path listed. There are two paths from
the switch to the storage for this adapter. For the next disk, set the priorities to 4, 1, 2, and 3.
In fail-over mode, assuming the workload is relatively balanced across the hdisks, this setting
balances the workload evenly across the paths.
Make sure that your system is equipped with the required file sets by running the lslpp
command as shown in Example 4-23.
Volume Groups
To avoid configuration problems and error log entries when you create Volume Groups using
iSCSI devices, follow these guidelines:
Configure Volume Groups that are created using iSCSI devices to be in an inactive state
after reboot. After the iSCSI devices are configured, manually activate the iSCSI-backed
Volume Groups. Then mount any associated file systems.
I/O failures
To avoid I/O failures, consider these recommendations:
If connectivity to iSCSI target devices is lost, I/O failures occur. Before doing anything that
causes long-term loss of connectivity to the active iSCSI targets, stop all I/O activity and
unmount iSCSI-backed file systems.
If a loss of connectivity occurs while applications are attempting I/O activities with iSCSI
devices, I/O errors eventually occur. You might not be able to unmount iSCSI-backed file
systems because the underlying iSCSI device remains busy.
File system maintenance must be performed if I/O failures occur due to loss of
connectivity to active iSCSI targets. To do file system maintenance, run the fsck command
against the affected file systems.
Note: A default initiator name is assigned when the software is installed. This
initiator name can be changed to match local network naming conventions.
You can also issue the lsattr command to verify the initiator_name parameter as
shown in Example 4-24.
f. The Maximum Targets Allowed field corresponds to the maximum number of iSCSI
targets that can be configured. If you reduce this number, you also reduce the amount
of network memory pre-allocated for the iSCSI protocol during configuration.
3. Right- click the new host name and select Add Port (Figure 4-3).
4. Configure the port as an iSCSI port and enter the IQN name that you collected in
Example 4-24. Add this value to the iSCSI Name as shown in Figure 4-4.
5. Create the LUNs in XIV and map them to the AIX iSCSI server so the server can see them
in the following steps.
7. The iSCSI connectivity panel in Figure 4-6 shows all the available iSCSI ports. Set the
MTU to 4500 if your network supports jumbo frames (Figure 4-6).
8. In the system view in the XIV GUI, right-click the XIV Storage box itself, and select
Properties Parameters.
9. Find the IQN of the XIV Storage System in the System Properties window (Figure 4-7).
If you are using XCLI, issue the config_get command as shown in Example 4-25.
10.Return to the AIX system and add the XIV iSCSI IP address, port name, and IQN to the
/etc/iscsi/targets file. This file must include the iSCSI targets for the device
configuration.
Tip: The iSCSI targets file defines the name and location of the iSCSI targets that the
iSCSI software initiator attempts to access. This file is read every time that the iSCSI
software initiator is loaded.
Each uncommented line in the file represents an iSCSI target. iSCSI device configuration
requires that the iSCSI targets can be reached through a properly configured network
interface. Although the iSCSI software initiator can work using a 10/100 Ethernet LAN, it is
designed for use with a separate gigabit Ethernet network.
Include your specific connection information in the targets file as shown in Example 4-26.
Example 4-26 Inserting connection information into the /etc/iscsi/targets file in AIX
# vi /etc/iscsi/targets
# the valid port is 5003
# the name of the target is iqn.com.ibm-4125-23WTT26
# The target line would look like:
# 192.168.3.2 5003 iqn.com.ibm-4125-23WWT26
#
# EXAMPLE 2: iSCSI Target with CHAP(MD5) authentication
# Assume the target is at address 10.2.1.105
# the valid port is 3260
# the name of the target is iqn.com.ibm-K167-42.fc1a
# the CHAP secret is "This is my password."
# The target line would look like:
# 10.2.1.105 3260 iqn.com.ibm-K167-42.fc1a "This is my password."
#
# EXAMPLE 3: iSCSI Target with CHAP(MD5) authentication and line continuation
# Assume the target is at address 10.2.1.106
# the valid port is 3260
# the name of the target is iqn.2003-01.com.ibm:00.fcd0ab21.shark128
# the CHAP secret is "123ismysecretpassword.fc1b"
# The target line would look like:
# 10.2.1.106 3260 iqn.2003-01.com.ibm:00.fcd0ab21.shark128 \
# "123ismysecretpassword.fc1b"
#
9.155.51.74 3260 iqn.2005-10.com.xivstorage:010114
Exception: If the appropriate disks are not defined, review the configuration of the
initiator, the target, and any iSCSI gateways to ensure correctness. Then rerun the
cfgmgr command.
The first step is to confirm that the network adapter supports jumbo frames. Jumbo frames
are Ethernet frames that support more than 1500 bytes. Jumbo frames can carry up to 9000
bytes of payload, but some care must be taken when using the term. Many different Gigabit
Ethernet switches and Gigabit Ethernet network cards can support jumbo frames. Check the
network card specification or the vendors support website to confirm if the network card
supports this functionality.
You can use lsattr to list some of the current adapter device driver settings. Enter lsattr -E
-l ent0 where ent0 is the adapter name. Make sure that you are checking and modifying the
correct adapter. A typical output is shown in Example 4-28.
In the example, jumbo_frames are off. With this setting not enabled, you are not able to
increase the network speed. Set up the tcp_sendspace, tcp_recvspace, sb_max, and mtu_size
network adapter and network interface options to optimal values.
To see the current settings, use lsattr to list the settings for tcp_sendspace, tcp_recvspace,
and mtu_size (Example 4-29).
Example 4-29 shows that all values are true and that mtu is set to 1500.
To change the mtu setting, enable jumbo_frames on the adapter. Issue the following
command:
chdev -l ent0 -a jumbo_frames=yes -P
Reboot the server by entering shutdown -Fr. Check the interface and adapter settings and
confirm the changes (Example 4-30).
Example 4-31 shows that the mtu value is changed to 9000. XIV supports only a mtu size of
4500.
Use the following command to change the mtu to 4500 on the AIX server:
chdev -l en0 -a mtu=4500
Confirm that the setting is changed. Use the /usr/sbin/no -a command to show the
sb_max, tcp_recvspace and tcp_sendspace values as shown in Example 4-32.
3. Map your volume to LUN 0, and it replaces the management LUN to your volume.
Non-XIV Devices
--------------------------
xiv_attach
The xiv_attach wizard is a utility that assists with attaching the server to the XIV system.
See Example 4-8 on page 148 to see the wizard and an example of what it does.
For more information, see the IBM Storage Host Software Solutions link in the XIV Infocenter
at:
http://publib.boulder.ibm.com/infocenter/ibmxiv/r2/index.jsp
When using AIX SAN boot in conjunction with XIV, the default MPIO is used. During the boot
sequence, AIX uses the bootlist to find valid paths to a LUN/hdisk that contains a valid boot
logical volume (hd5). However, a maximum of five paths can be defined in the bootlist, while
the XIV multipathing setup results in more than five paths to a hdisk. A fully redundant
configuration establishes 12 paths (Figure 1-7 on page 15).
For example, consider two hdisks (hdisk0 and hdisk1) containing a valid boot logical volume,
both having 12 paths to the XIV Storage System. To set the bootlist for hdisk0 and hdisk1,
issue the following command:
/ > bootlist -m normal hdisk0 hdisk1
Example 4-35 shows that hdisk1 is not present in the bootlist. Therefore, the system cannot
boot from hdisk1 if the paths to hdisk0 are lost.
There is a workaround in AIX 6.1 TL06 and AIX 7.1 to control the bootlist using the pathid
parameter as in the following command:
bootlist m normal hdisk0 pathid=0 hdisk0 pathid=1 hdisk1 pathid=0 hdisk1 pathid=1
Implement SAN boot with AIX using one of the following methods:
For a system with an already installed AIX operating system, mirror the rootvg volume to
the SAN disk.
For a new system, start the AIX installation from a bootable AIX CD installation package or
use Network Installation Management (NIM).
The method known as mirroring is simpler to implement than using the Network Installation
Manager.
To create a boot disk on the XIV system, perform the following steps:
1. Select a logical drive that is the same size or larger than the size of rootvg that are
currently on the internal SCSI disk. Verify that your AIX system can see the new disk with
the lspv -L command as shown in Example 4-36.
2. Verify the size with the xiv_devlist command to make sure that you are using an XIV
(external) disk. Example 4-37 shows that hdisk0 is 32 GB, hdisks 1 through 5 are
attached, and they are XIV LUNs. Notice that hdisk1 is only 17 GB, so it is not large
enough to create a mirror.
XIV Devices
-----------------------------------------------------------------------------
Device Size (GB) Paths Vol Name Vol Id XIV Id XIV Host
-----------------------------------------------------------------------------
Non-XIV Devices
-----------------------------
Device Size (GB) Paths
-----------------------------
/dev/hdisk0 32.2 1/1
-----------------------------
3. Add the new disk to the rootvg volume group by clicking smitty vg Set Characteristics
of a Volume Group Add a Physical Volume to a Volume Group.
4. Leave Force the creation of volume group set to no.
5. Enter the Volume Group name (in this example, rootvg) and Physical Volume name
that you want to add to the volume group (Figure 4-9).
6. Create the mirror of rootvg. If the rootvg is already mirrored, create a third copy on the
new disk by clicking smitty vg Mirror a Volume Group (Figure 4-11).
Enter the volume group name that you want to mirror (rootvg, in this example).
7. Select the one of the following mirror sync modes:
Foreground: This option causes the command to run until the mirror copy
synchronization completes. The synchronization can take a long time. The amount of
time depends mainly on the speed of your network and how much data you have.
Background: This option causes the command to complete immediately, and mirror
copy synchronization occurs in the background. With this option, it is not obvious when
the mirrors complete their synchronization.
No Sync: THis option causes the command to complete immediately without
performing any type of mirror synchronization. If this option is used, the new remote
mirror copy exists but is marked as stale until it is synchronized with the syncvg
command.
8. Select the Physical Volume name. You added this drive to your disk group in Figure 4-9 on
page 170. The number of copies of each logical volume is the number of physical
partitions allocated for each logical partition. The value can be one to three. A value of two
or three indicated a mirrored logical volume. Leave the Keep Quorum Checking on and
Create Exact LV Mapping settings set to no.
9. Verify that all partitions are mirrored with lsvg -l rootvg (Figure 4-13). The Physical
Volume (PVs) column displays as two or three, depending on the number you chose when
you created the mirror.
10.Recreate the boot logical drive, and change the normal boot list with the following
commands:
bosboot -ad hdiskx
bootlist -m normal hdiskx
Figure 4-14 shows the output after running the commands.
11.Select the rootvg volume group and the original hdisk that you want to remove, then click
smitty vg Unmirror a Volume Group.
12.Select rootvg for the volume group name ROOTVG and the internal SCSI disk you want to
remove.
13.Click smitty vg Set Characteristics of a Volume Group Remove a Physical
Volume from a Volume Group.
At this stage, the creation of a bootable disk on the XIV is completed. Restarting the system
makes it boot from the SAN (XIV) disk.
After the system reboots, use the lspv -L command to confirm that the server is booting from
the XIV hdisk as shown in Figure 4-15.
Tip: If the system cannot see the SAN fabric at login, configure the HBAs at the server
open firmware prompt.
Because a SAN allows access to many devices, identifying the hdisk to install to can be
difficult. Use the following method to facilitate the discovery of the lun_id to hdisk correlation:
1. If possible, zone the switch or disk array such that the system being installed can discover
only the disks to be installed to. After the installation completes, you can reopen the
zoning so the system can discover all necessary devices.
The Open Firmware implementation can only boot from lun_ids 0 through 7. The firmware on
the Fibre Channel adapter (HBA) promotes this lun_id to an 8-byte FC lun-id. The firmware
does this promotion by adding a byte of zeros to the front and 6 bytes of zeros to the end. For
example, lun_id 2 becomes 0x0002000000000000. The lun_id is normally displayed without
the leading zeros. Take care when installing because the procedure allows installation to
lun_ids outside of this range.
Installation procedure
To install on external storage, follow these steps:
1. Insert an AIX CD that has a bootable image into the CD-ROM drive.
2. Select CD-ROM as the installation device to make the system boot from the CD. The way
to change the bootlist varies model by model. In most System p models, you use the
System Management Services (SMS) menu. For more information, see the users guide
for your model.
3. Allows the system to boot from the AIX CD image after you leave the SMS menu.
4. After a few minutes, the console displays a window that directs you to press a key to use
the device as the system console.
5. A window prompts you to select an installation language.
6. The Welcome to the Base Operating System Installation and Maintenance window is
displayed. Change the installation and system settings for this system to select a Fibre
Channel-attached disk as a target disk. Enter 2 to continue.
7. On the Installation and Settings window, enter 1 to change the system settings and select
the New and Complete Overwrite option.
8. On the Change (the destination) Disk window, select the Fibre Channel disks that are
mapped to your system. To see more information, enter 77 to display the detailed
information window that includes the PVID. Enter 77 again to show WWPN and LUN_ID
Important: Verify that you made the correct selection for root volume group. The
existing data in the destination root volume group is deleted during Base Operating
System (BOS) installation.
When the system reboots, a window displays the address of the device from which the
system is reading the boot image.
Deploy the NIM environment, and make the following configurations on the NIM master:
The NIM server is properly configured as the NIM master and the basic NIM resources are
defined.
The Fibre Channel adapters are already installed on the system onto which AIX is to be
installed.
The Fibre Channel adapters are connected to a SAN, and on the XIV system have at least
one logical volume (LUN) mapped to the host.
The target system (NIM client) currently has no operating system installed and is
configured to boot from the NIM server.
For more information about how to configure a NIM server, see the AIX 5L Version 5.3:
Installing AIX reference, SC23-4887-02.
Installation procedure
Before the installation, modify the bosinst.data file, where the installation control is stored.
Insert your appropriate values at the following stanza:
SAN_DISKID
This stanza specifies the worldwide port name and a logical unit ID for Fibre
Channel-attached disks. The worldwide port name and logical unit ID are in the format
returned by the lsattr command (that is, 0x followed by 116 hexadecimal digits). The
ww_name and lun_id are separated by two slashes (//).
SAN_DISKID = <worldwide_portname//lun_id>
For example:
SAN_DISKID = 0x0123456789FEDCBA//0x2000000000000
For the latest information, see the Host Attachment Kit publications at:
http://publib.boulder.ibm.com/infocenter/ibmxiv/r2/topic/com.ibm.help.xiv.doc/xiv_
pubsrelatedinfoic.html
Important: The procedures and instructions given here are based on code that was
available at the time of writing this book. For the latest support information and instructions,
see the System Storage Interoperability Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
The HP-UX host attachment process with XIV is described in the Host Attachment Guide for
HPUX, which is available at the IBM XIV Information Center:
http://publib.boulder.ibm.com/infocenter/ibmxiv/r2/topic/com.ibm.help.xiv.doc/xiv_
pubsrelatedinfoic.html
This section focuses on the HP-UX specific steps. The steps that are not specific to HP-UX
are described in Chapter 1, Host connectivity on page 1.
Figure 5-1 shows the host object that was defined for the HP-UX server used for the
examples in this book.
Figure 5-2 shows the volumes that were defined for the HP-UX server.
Vendor ID is = 0x1077
Device ID is = 0x2312
PCI Sub-system Vendor ID is = 0x103C
PCI Sub-system ID is = 0x12BA
PCI Mode = PCI-X 133 MHz
ISP Code version = 3.3.166
ISP Chip version = 3
Topology = PTTOPT_FABRIC
Link Speed = 2Gb
Local N_Port_id is = 0x0b3400
Previous N_Port_id is = None
N_Port Node World Wide Name = 0x50060b000039dde1
N_Port Port World Wide Name = 0x50060b000039dde0
Switch Port World Wide Name = 0x203200053353e557
Switch Node World Wide Name = 0x100000053353e557
Driver state = ONLINE
Hardware Path is = 0/2/1/0
Maximum Frame Size = 2048
Driver-Firmware Dump Available = NO
Driver-Firmware Dump Timestamp = N/A
Driver Version = @(#) fcd B.11.31.01 Jan 7 2007
The XIV Host Attachment Kit version 1.7.0 supports HP-UX 11iv3 on HP Integrity servers. For
attachment of HP-UX 11iv2 or HP-UX 11iv3 on PA-RISC, use Host Attachment Kit version
1.6.0. The Host Attachment Kit includes scripts to facilitate HP-UX attachment to XIV. For
example, the xiv_attach script performs the following tasks (Example 5-2):
Identifies the Fibre Channel adapters of the hosts connected to XIV storage systems.
Identifies the name of the host object defined on the XIV storage system for this host (if
already created).
Supports rescanning for new storage devices.
-------------------------------------------------------------------------------
Only fibre-channel is supported on this host.
Would you like to set up an FC attachment? [default: yes ]:
-------------------------------------------------------------------------------
Please wait while the wizard validates your existing configuration...
This host is already configured for the XIV system
-------------------------------------------------------------------------------
Please zone this host and add its WWPNs with the XIV storage system:
50060b000039dde0: /dev/fcd0: []:
50060b000039dde2: /dev/fcd1: []:
500110a000101960: /dev/fcd2: []:
500110a000101962: /dev/fcd3: []:
Press [ENTER] to proceed.
Would you like to rescan for new storage devices now? [default: yes ]:
Please wait while rescanning for storage devices...
-------------------------------------------------------------------------------
The host is connected to the following XIV storage arrays:
Serial Version Host Defined Ports Defined Protocol Host Name(s)
1310114 11.0.0.0 Yes All FC ITSO_HP-UX
This host is defined on all FC-attached XIV storage arrays
-------------------------------------------------------------------------------
The XIV host attachment wizard successfully configured this host
HP introduced HP Native Multi-Pathing with HP-UX 11iv3. The earlier pvlinks multi-pathing is
still available, but use Native Multi-Pathing. HP Native Multi-Pathing provides I/O load
balancing across the available I/O paths, whereas pvlinks provides path failover and failback,
but no load balancing. Both multi-pathing methods can be used for HP-UX attachment to XIV.
HP Native Multi-Pathing uses the so-called Agile View Device Addressing which addresses a
device by its worldwide ID (WWID) as an object. The device can be discovered by its WWID
regardless of the hardware controllers, adapters, or paths between the HP-UX server and the
device itself. Therefore, this addressing method creates only one device file for each device.
Example 5-3 shows the HP-UX view of five XIV volumes using agile addressing and the
conversion from agile to legacy view.
If device special files are missing on the HP-UX server, you can create them in two ways. The
first option is rebooting the host, which is disruptive. The other option is to run the command
insf -eC disk, which will reinstall the special device files for all devices of the class disk.
Volume groups, logical volumes, and file systems can be created on the HP-UX host.
Example 5-4 shows the HP-UX commands to initialize the physical volumes and create a
volume group in an LVM environment. The rest is standard HP-UX system administration, not
specific to XIV and therefore is not addressed.
HP Native Multi-Pathing is used to automatically specify the Agile View device files, for
example /dev/(r)disk/disk1299. To use pvlinks, specify the Legacy View device files of all
available hardware paths to a disk device, for example /dev/(r)dsk/c153t0d1 and c155t0d1.
According to the HP-UX System Administrator's Guide, both volume managers can coexist on
an HP-UX server. For more information, see HP-UX System Administration on the following
page:
http://www.hp.com/go/hpux-core-docs
You can use both simultaneously (on separate physical disks), but usually you choose to use
one or the other exclusively.
The configuration of XIV volumes on HP-UX with LVM is described in 5.2, HP-UX
multi-pathing solutions on page 180. Example 5-5 shows the initialization of disks for VxVM
use and the creation of a disk group with the vxdiskadm utility.
Example 5-5 Disk initialization and disk group creation with vxdiskadm
# vxdctl enable
# vxdisk list
DEVICE TYPE DISK GROUP STATUS
Disk_0s2 auto:LVM - - LVM
Disk_1s2 auto:LVM - - LVM
XIV2_0 auto:none - - online invalid
XIV2_1 auto:none - - online invalid
XIV2_2 auto:none - - online invalid
XIV2_3 auto:LVM - - LVM
XIV2_4 auto:LVM - - LVM
# vxdiskadm
Use this operation to add one or more disks to a disk group. You can
add the selected disks to an existing disk group or to a new disk group
that will be created as a part of the operation. The selected disks may
also be added to a disk group as spares. Or they may be added as
nohotuses to be excluded from hot-relocation use. The selected
disks may also be initialized without adding them to a disk group
leaving the disks available for use as replacement disks.
More than one disk or pattern may be entered at the prompt. Here are
some disk selection examples:
XIV2_1 XIV2_2
A new disk group will be created named dg01 and the selected disks
will be added to the disk group with default disk names.
XIV2_1 XIV2_2
Do you want to use the default layout for all disks being initialized?
[y,n,q,?] (default: y) n
Do you want to use the same layout for all disks being initialized?
[y,n,q,?] (default: y)
# vxdisk list
DEVICE TYPE DISK GROUP STATUS
Disk_0s2 auto:LVM - - LVM
Disk_1s2 auto:LVM - - LVM
XIV2_0 auto:none - - online invalid
XIV2_1 auto:hpdisk dg0101 dg01 online
XIV2_2 auto:hpdisk dg0102 dg01 online
XIV2_3 auto:LVM - - LVM
XIV2_4 auto:LVM - - LVM
In this example, after having creating the disk groups and the VxVM disks, you must create
file systems and mount them.
On a host system, the VxVM command vxddladm listsupport displays a list of storage
systems supported by the VxVM version installed on the operating system (Example 5-6).
On a host system, ASLs allow easier identification of the attached disk storage devices. The
ASL serially numbers the attached storage systems of the same type and the volumes of a
single storage system being assigned to this host.
Example 5-7 shows that five volumes of one XIV system are assigned to that HP-UX host.
VxVM controls the devices XIV2_1 and XIV2_2, and the disk group name is dg01. The HP
LVM controls the remaining XIV devices, except for XIV2_0.
ASL packages for XIV and HP-UX 11iv3 are available for download at:
http://www.symantec.com/business/support/index?page=content&id=TECH63130
Installing HP-UX
The examples in this chapter involve an HP-UX installation on HP Itanium-based Integrity
systems. On older HP PA-RISC systems, the processes to boot the server and select disks to
install HP-UX to are different. A complete description of the HP-UX installation processes on
HP Integrity and PA-RISC systems is provided in the HP manual HP-UX 11iv3 Installation and
Update Guide. Click HP-UX 11iv3 at:
http://www.hp.com/go/hpux-core-docs
To install HP-UX 11iv3 on an XIV volume from DVD - on an HP Integrity system, perform
these steps:
1. Insert the first HP-UX Operating Environment DVD into the DVD drive.
2. Reboot or power on the system and wait for the EFI panel.
4. The server boots from the installation media. On the HP-UX installation and recovery
process window, select Install HP-UX (Figure 5-5).
6. The remaining steps of an HP-UX installation on a SAN disk do not differ from installation
on an internal disk.
The storage-specific part is the identification of the XIV volume to install to on HP-UX. For
more information, see 5.4.1, Installing HP-UX on external storage on page 187.
Important: The procedures and instructions given here are based on code that was
available at the time of writing this book. For the latest support information and instructions,
ALWAYS see the System Storage Interoperability Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
Tip: You can use both Fibre Channel and iSCSI connections to attach hosts. However, do
not use both connections for the same LUN on the same host.
The environment in the examples presented in this chapter consists of a SUN Sparc T5220
running with Solaris 10 U10.
3. Change to the newly created directory and start the Host Attachment Kit installer as
shown in Example 6-3.
Tip: You must be logged in as root or have root privileges to use the Host Attachment Kit.
The main executable files are installed in the folder seen in Example 6-4.
To configure your system in and for the XIV, you need to set up your SAN zoning first so that
the XIV is visible for the Host. To start the configuration, perform the following steps:
1. Run the xiv_attach command, which is mandatory for support. Example 6-5 shows an
example of host configuration using the command.
Remember: After running the xiv_attach command for the first time, the server will
need to be rebooted.
The wizard will now validate host configuration for the XIV system.
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
Please choose a connectivity type, [f]c / [i]scsi : f
-------------------------------------------------------------------------------
Please wait while the wizard validates your existing configuration...
The wizard needs to configure the host for the XIV system.
Do you want to proceed? [default: yes ]: yes
Please wait while the host is being configured...
devfsadm: driver failed to attach: sgen
2. After the system reboot, start xiv_attach again to finish the system to XIV configuration
for the Solaris host as seen in Example 6-6.
The wizard will now validate host configuration for the XIV system.
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
Please choose a connectivity type, [f]c / [i]scsi : f
-------------------------------------------------------------------------------
Please wait while the wizard validates your existing configuration...
The wizard needs to configure the host for the XIV system.
Do you want to proceed? [default: yes ]: yes
Please wait while the host is being configured...
The host is now being configured for the XIV system
-------------------------------------------------------------------------------
Please zone this host and add its WWPNs with the XIV storage system:
2101001b32b1d4b1: /dev/cfg/c5: [QLogic Corp.]: 371-4325-01
2100001b3291d4b1: /dev/cfg/c4: [QLogic Corp.]: 371-4325-01
Press [ENTER] to proceed.
Would you like to rescan for new storage devices now? [default: yes ]: yes
Please wait while rescanning for storage devices...
-------------------------------------------------------------------------------
The host is connected to the following XIV storage arrays:
Serial Version Host Defined Ports Defined Protocol Host Name(s)
1310114 0000 No None FC --
1310133 0000 No None FC --
1300203 10.2 No None FC --
This host is not defined on some of the FC-attached XIV storage systems.
Do you wish to define this host on these systems now? [default: yes ]: yes
Please enter a name for this host [default: sun-t5220 ]:
Please enter a username for system 1310114 : [default: admin ]: itso
Please enter the password of user itso for system 1310114:
xiv_attach detected connectivity to three XIVs (zoning to XIVs is already completed) and
checked whether a valid host definition exists.
3. If it does not exist, choose whether you want to have hosts defined within each XIV.
4. Provide a user (with storageadmin rights) and password for each detected XIV. It then
connects to each remote XIV and defines the host and ports on the XIV. Example 6-7
shows the newly created host output from one of the XIVs
Tip: A rescan of for new XIV LUNs can be done with xiv_fc_admin -R.
The wizard will now validate host configuration for the XIV system.
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
The XIV host attachment wizard successfully configured this host
2. Define the host and iSCSI port when it rescans for storage devices as seen in
Example 6-9 on page 195. You need a valid storageadmin ID to do so. The host and iSCSI
port can also be defined on the GUI.
3. Discover the iSCSI qualified name (IQN) of your server with the xiv_iscsi_admin -P
command as shown in Example 6-10.
Options:
-h, --help show this help message and exit
-t OUT, --out=OUT Choose output method: tui, csv, xml (default: tui)
-o FIELDS, --options=FIELDS
Fields to display; Comma-separated, no spaces. Use -l
to see the list of fields
-f OUTFILE, --file=OUTFILE
File to output to (instead of STDOUT) - can be used
only with -t csv/xml
-H, --hex Display XIV volume and machine IDs in hexadecimal base
-u SIZE_UNIT, --size-unit=SIZE_UNIT
The size unit to use (e.g. MB, GB, TB, MiB, GiB, TiB,
...)
-d, --debug Enable debug logging
-l, --list-fields List available fields for the -o option
-m MP_FRAMEWORK_STR, --multipath=MP_FRAMEWORK_STR
Enforce a multipathing framework <auto|native|veritas>
-x, --xiv-only Print only XIV devices
-V, --version Shows the version of the HostAttachmentKit framework
xiv_diag
This utility gathers diagnostic information from the operating system. The resulting
compressed file can then be sent to IBM-XIV support teams for review and analysis.
Example results are shown in Example 6-13.
FORMAT MENU:
disk - select a disk
type - select (define) a disk type
partition - select (define) a partition table
current - describe the current disk
format - format and analyze the disk
repair - repair a defective sector
label - write label to the disk
analyze - surface analysis
defect - defect list management
backup - search for backup labels
verify - read and display labels
inquiry - show vendor, product and revision
volname - set 8-character volume name
!<cmd> - execute <cmd>, then return
quit
The standard partition table can be used, but you can also define a user-specific table. Use
the partition command in the format tool to change the partition table. You can see the
newly defined table using the print command as shown in Example 6-16.
partition> label
Ready to label disk, continue? yes
partition> quit
format> quit
After mounting the volume as shown in Example 6-20, you can start using the volume with ufs
file systems.
Important: The procedures and instructions given here are based on code that was
available at the time of writing this book. For the latest support information and instructions,
see the System Storage Interoperability Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
At the time of writing, XIV supports the use of VxVM and DMP with the following operating
systems:
HP-UX
AIX
Redhat Enterprise Linux
SUSE Linux
Linux on Power
Solaris
Depending on the OS version and hardware platform, only specific versions and releases of
Veritas Volume Manager are supported when connecting to XIV. In general, IBM supports
VxVM versions 5.0, and 5.1.
For most of the OS and VxVM versions, IBM supports space reclamation on thin provisioned
volumes.
For more information about the operating systems and VxVM versions supported, see the
System Storage Interoperability Center at:
http://www.ibm.com/systems/support/storage/config/ssic
In addition, you can find information about attaching the IBM XIV Storage System to hosts
with VxVM and DMP at the Symantec website:
https://sort.symantec.com/asl
7.2 Prerequisites
Common prerequisites such as cabling, SAN zoning defined, and volumes being created and
mapped to the host, must be completed. In addition, the following tasks must be completed to
successfully attach XIV to host systems using VxVM with DMP:
Check Array Support Library (ASL) availability for XIV Storage System on your Symantec
Storage Foundation installation.
Place the XIV volumes under VxVM control.
Set up DMP multipathing with IBM XIV.
Make sure that you install all the patches and updates available for your Symantec Storage
Foundation installation. For more information, see your Symantec Storage Foundation
documentation.
Example 7-1 Checking the availability ASL for IBM XIV Storage System
# vxddladm listversion|grep xiv
libvxxiv.so vm-5.1.100-rev-1 5.1
If the command output does not show that the required ASL is already installed, locate the
installation package. The installation package for the ASL is available at:
https://sort.symantec.com/asl
Specify the vendor of your storage system, your operating system, and the version of your
Symantec Storage Foundation. You are then redirected to a page from which you can
download the ASL package for your environment. Installation instructions are available on the
same page.
Install the required XIV Host Attachment Kit for your platform. You can check the Host
Attachment Kit availability for your platform at:
http://www.ibm.com/support/fixcentral
3. Change to the newly created directory and start the Host Attachment Kit installer, as seen
in Example 7-3.
Note: You must be logged in as root or have root privileges to use the Host Attachment Kit.
The wizard will now validate host configuration for the XIV system.
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
Please choose a connectivity type, [f]c / [i]scsi : f
Notice: VxDMP is available and will be used as the DMP software
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
Please wait while the wizard validates your existing configuration...
The wizard needs to configure the host for the XIV system.
Do you want to proceed? [default: yes ]:
Please wait while the host is being configured...
-------------------------------------------------------------------------------
A reboot is required in order to continue.
Please reboot the machine and restart the wizard
Press [ENTER] to exit.
2. For the Solaris on SUN server used in this example, you must reboot the host before
proceeding to the next step. Other systems can vary.
3. After the system reboot, start xiv_attach again to complete the host system configuration
for XIV attachment (Example 7-5).
4. Create the host on XIV and map volumes (LUNs) to the host system. You can use the XIV
GUI for that task, as illustrated in 1.4, Logical configuration for host connectivity on
page 35.
5. After the LUN mapping is completed, discover the mapped LUNs on your host by running
the xiv_fc_admin -R command.
6. Use the command /opt/xiv/host_attach/bin/xiv_devlist to check the mapped
volumes and the number of paths to the XIV Storage System (Example 7-6).
More than one disk or pattern may be entered at the prompt. Here are
xiv0_03d2 xiv0_03d3
xiv0_03d2 xiv0_03d3
Continue with operation? [y,n,q,?] (default: y) y
The following disk devices have a valid VTOC, but do not appear to have
been initialized for the Volume Manager. If there is data on the disks
that should NOT be destroyed you should encapsulate the existing disk
partitions as volumes instead of adding the disks as new disks.
Output format: [Device_Name]
xiv0_03d2 xiv0_03d3
Do you want to use the default layout for all disks being initialized?
[y,n,q,?] (default: y)
Tip: If the vxdiskadm initialization function complains the disk is offline, you might need
to initialize it using the default OS-specific utility. For example, use the format command
in Solaris.
5. Check the results using the vxdisk list and vxdg list commands as shown in
Example 7-9.
Example 7-9 Showing the results of putting XI V LUNs under VxVM control
# vxdisk list
DEVICE TYPE DISK GROUP STATUS
disk_0 auto:ZFS - - ZFS
disk_1 auto:ZFS - - ZFS
disk_2 auto:ZFS - - ZFS
disk_3 auto:ZFS - - ZFS
xiv0_03d2 auto:cdsdisk XIV_DG01 XIV_DG online thinrclm
xiv0_03d3 auto:cdsdisk XIV_DG02 XIV_DG online thinrclm
# vxdg list
NAME STATE ID
XIV_DG enabled,cds 1316177502.13.sun-t5220-02-1
# vxdg list XIV_DG
Group: XIV_DG
dgid: 1316177502.13.sun-t5220-02-1
import-id: 1024.12
flags: cds
version: 160
alignment: 8192 (bytes)
ssb: on
autotagging: on
detach-policy: global
dg-fail-policy: dgdisable
copies: nconfig=default nlog=default
config: seqno=0.1029 permlen=48144 free=48140 templen=2 loglen=7296
config disk xiv0_03d2 copy 1 len=48144 state=clean online
config disk xiv0_03d3 copy 1 len=48144 state=clean online
log disk xiv0_03d2 copy 1 len=7296
log disk xiv0_03d3 copy 1 len=7296
The XIV LUNs that were added are now available for volume creation and data storage.
The status thinrclm means that the volumes from the XIV are thin-provisioned, and the
XIV storage has the Veritas thin reclamation API implemented.
6. Use the vxdisk reclaim <diskgroup> | <disk> command to free up any space that can
be reclaimed.
Example 7-10 Identifying names of enclosures seated on an IBM XIV Storage System
# vxdmpadm listenclosure all
ENCLR_NAME ENCLR_TYPE ENCLR_SNO STATUS ARRAY_TYPE LUN_COUNT
==================================================================================
disk Disk DISKS CONNECTED Disk 4
xiv0 XIV 130210114 CONNECTED A/A 2
3. Change the iopolicy parameter for the identified enclosures by executing the command
vxdmpadm setattr enclosure <identified enclosure name> iopolicy=round-robin for
each identified enclosure.
4. Check the results of the change by executing the command vxdmpadm getattr
enclosure <identified enclosure name> as shown in Example 7-11.
5. For heavy workloads, increase the queue depth parameter to 64. The queue depth can be
set as high as 128 if needed. Run the command vxdmpadm gettune dmp_queue_depth
to get information about current settings and run vxdmpadm settune
dmp_queue_depth=<new queue depth value> to adjust them as shown in
Example 7-12.
Whenever a disk is brought online, the current UDID value is compared to the UDID stored in
the private region of the disk. If the UDID does not match, the udid_mismatch flag is set on
the disk. This flag allows LUN snapshots to be imported on the same host as the original
LUN. It also allows multiple snapshots of the same LUN to be concurrently imported on a
single server. These snapshots can then be used for the offline backup or processing.
After creating XIV snapshots for LUNs used on a host under VxVM control, unlock (enable
writing) those snapshots and map them to your host. When this process is complete, the
snapshot LUNS can be imported on the host using the following steps:
1. Check that the created snapshots are visible for your host by running the commands
vxdisk scandisks and vxdisk list as shown in Example 7-13.
2. Import the created snapshot on your host by running the command vxdg -n <name for
new volume group> -o useclonedev=on,updateid -C import <name of original
volume group>.
3. Run the vxdisk list command to ensure that the LUNs were imported as shown in
Example 7-14.
Important: The procedures and instructions given here are based on code that was
available at the time of writing this book. For the latest support information and instructions,
see the System Storage Interoperability Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
PowerVM offers a secure virtualization environment with the following features and benefits:
Consolidates diverse sets of applications that are built for multiple operating systems (AIX,
IBM i, and Linux) on a single server
Virtualizes processor, memory, and I/O resources to increase asset utilization and reduce
infrastructure costs
Dynamically adjusts server capability to meet changing workload demands
Moves running workloads between servers to maximize availability and avoid planned
downtime
With PowerVM Express Edition, you can create up to three partitions on a server (two client
partitions and one for the VIOS and IVM). You can use virtualized disk and optical devices,
With PowerVM Standard Edition, you can create up to 254 partitions on a server. You can
use virtualized disk and optical devices, and try out the shared processor pool. All
virtualization features can be managed using a Hardware Management Console or the IVM.
These features include Micro-Partitioning, shared processor pool, Virtual I/O Server,
PowerVM Lx86, shared dedicated capacity, NPIV, and virtual tape.
VIOS owns physical I/O resources such as Ethernet and SCSI/FC adapters. It virtualizes
those resources for its client LPARs to share them remotely using the built-in hypervisor
services. These client LPARs can be created quickly, and typically own only real memory and
shares of processors without any physical disks or physical Ethernet adapters.
With Virtual SCSI support, VIOS client partitions can share disk storage that is physically
assigned to the VIOS LPAR. This virtual SCSI support of VIOS can be used with storage
devices that do not support the IBM i proprietary 520-byte/sectors format available to IBM i
clients of VIOS. These storage devices include IBM XIV Storage System server.
VIOS owns the physical adapters, such as the Fibre Channel storage adapters, that are
connected to the XIV system. The logical unit numbers (LUNs) of the physical storage
devices that are detected by VIOS are mapped to VIOS virtual SCSI (VSCSI) server
adapters. The VSCSI adapters are created as part of its partition profile.
Figure 8-1 shows an example of the VIOS owning the physical disk devices and their virtual
SCSI connections to two client partitions.
SCSI SCSI
h d is k h d is k
LUNs LUNs
#1 #n
... # 1 -(m -1 ) # m -n
P O W E R H y p e rv is o r
F C a d a p te r F C a d a p te r
X IV S to ra g e S y s te m
The VIOS does not do any device discovery on ports using NPIV. Therefore, no devices are
shown in the VIOS connected to NPIV adapters. The discovery is left for the virtual client, and
all the devices found during discovery are detected only by the virtual client. This way, the
virtual client can use FC SAN storage-specific multipathing software on the client to discover
and manage devices.
For more information about PowerVM virtualization management, see IBM PowerVM
Virtualization Managing and Monitoring, SG24-7590.
Important: The procedures and instructions given here are based on code that was
available at the time of writing this book. For the latest support information and instructions,
see the System Storage Interoperability Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
8.2.1 Requirements
The following are general requirements, current at the time of writing, to fulfill when attaching
an XIV Storage System to an IBM i VIOS client (Table 8-1).
These requirements serve as an orientation to the required hardware and software levels for
XIV Storage system with IBM i. For current information, see the System Storage
Interoperation Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
Important: When the IBM i operating system and VIOS are on an IBM Power Blade
server, you can define only one VSCSI adapter to assign to an IBM i client. Therefore, the
number of LUNs that can connect to the IBM i operating system is limited to 16.
8.2.4 Queue depth in the IBM i operating system and Virtual I/O Server
When connecting the IBM XIV Storage System server to an IBM i client through the VIOS,
consider the following types of queue depths:
The IBM i queue depth to a virtual LUN
SCSI command tag queuing in the IBM i operating system allows up to 32 I/O operations
to one LUN at the same time.
The IBM i operating system has a fixed queue depth of 32, which is not changeable. However,
the queue depths in the VIOS can be set up by a user. The default setting in the VIOS varies
based on the type of connected storage, type of physical adapter, and type of multipath driver
or Host Attachment Kit.
Check the queue depth on physical disks by entering the following VIOS command:
lsdev -dev hdiskxx -attr queue_depth
This last command ensures that the queue depth in the VIOS matches the IBM i queue depth
for an XIV LUN.
For more information, see 8.3.3, IBM i multipath capability with two Virtual I/O Servers on
page 226.
With the grid architecture and massive parallelism inherent to XIV system, the general
approach is to maximize the use of all XIV resources at all times.
Distributing connectivity
The goal for host connectivity is to create a balance of the resources in the IBM XIV Storage
System server. Balance is achieved by distributing the physical connections across the
interface modules. A host usually manages multiple physical connections to the storage
device for redundancy purposes using a SAN connected switch. The ideal is to distribute
You do not need to connect each host instance to each interface module. However, when the
host has more than one physical connection, have the connections (cabling) spread across
separate interface modules.
Similarly, if multiple hosts have multiple connections, you must distribute the connections
evenly across the interface modules.
Appropriate zoning: Use a separate zone for each host adapter (initiator). For each zone
that contains the host adapter, add all switch port connections from the XIV Storage
System (targets).
Queue depth
SCSI command tag queuing for LUNs on the IBM XIV Storage System server allows multiple
I/O operations to one LUN at the same time. The LUN queue depth indicates the number of
I/O operations that can be done simultaneously to a LUN.
The XIV architecture eliminates the storage concept of a large central cache. Instead, each
module in the XIV grid has its own dedicated cache. The XIV algorithms that stage data
between disk and cache work most efficiently when multiple I/O requests are coming in
parallel. This process is where the host queue depth becomes an important factor in
maximizing XIV I/O performance. Therefore, configure the host HBA queue depths as large
as possible.
11.In the Alternate Restart Device window, if the virtual DVD-RAM device is used in the IBM i
client, select the corresponding virtual adapter.
12.In the Console Selection window, select the default of HMC for the console device and
click OK.
2. Configure TCP/IP for the logical Ethernet adapter entX using the mktcpip command
syntax. Specify the corresponding interface resource enX.
3. Verify the created TCP/IP connection by pinging the IP address that you specified in the
mktcpip command.
With Virtual I/O Server release 2.1.2 or later, and IBM i release 6.1.1 or later, you can
establish multipath to a set of LUNs. Each path uses a connection through a separate VIOS.
This topology provides redundancy if a connection or the VIOS fails. Up to eight multipath
connections can be implemented to the same set of LUNs, each through a separate VIOS.
However, most IT centers establish no more than two such connections.
For testing purposes, separate switches were not used. Instead, separate blades in the same
SAN Director were used. In a real, production environment, use separate switches for
redundancy as shown in Figure 8-5.
POWER6
7 3 16 16
Physical FC adapters
Hypervisor
Switches
XIV
XIV LUNs
To connect XIV LUNs to an IBM i client partition in multipath with two VIOS, perform these
steps:
1. After the LUNs are created in the XIV system, map the LUNs to the VIOS host as shown in
8.4, Mapping XIV volumes in the Virtual I/O Server on page 229. You can use the XIV
Storage Management GUI or Extended Command Line Interface (XCLI).
2. Log on to VIOS as administrator. The example uses PuTTY to log in as described in 6.5,
Configuring VIOS virtual devices, in IBM i and Midrange External Storage, SG24-7668.
Run the cfgdev command so that the VIOS can recognize newly attached LUNs.
As previously explained, for a multipath setup for IBM i, each XIV LUN is connected to
both VIOS partitions. Before assigning these LUNs (from any of the VIOS partitions) to the
IBM i client, make sure that the volume is not SCSI reserved.
b. From the HMC, select the partition and choose Configuration Manage Profiles.
c. Select the profile and click Actions Edit.
5. Map the relevant hdisks to the VSCSI adapter by entering the following command:
mkvdev -vdev hdiskx -vadapter <name>
In this example, the XIV LUNs are mapped to the adapter vhost5. Each LUN is given a
virtual device name using the -dev parameter as shown in Figure 8-10.
After completing these steps for each VIOS, the LUNs are available to the IBM i client in
multipath (one path through each VIOS).
XIV Devices
-------------------------------------------------------------------------
Device Size Paths Vol Name Vol Id XIV Id XIV Host
-------------------------------------------------------------------------
/dev/hdisk5 154.6GB 2/2 ITSO_i_1 4353 1300203 VIOS_1
-------------------------------------------------------------------------
/dev/hdisk6 154.6GB 2/2 ITSO_i_CG.snap 4497 1300203 VIOS_1
_group_00001.I
TSO_i_4
-------------------------------------------------------------------------
/dev/hdisk7 154.6GB 2/2 ITSO_i_3 4355 1300203 VIOS_1
-------------------------------------------------------------------------
/dev/hdisk8 154.6GB 2/2 ITSO_i_CG.snap 4499 1300203 VIOS_1
_group_00001.I
TSO_i_6
-------------------------------------------------------------------------
/dev/hdisk9 154.6GB 2/2 ITSO_i_5 4357 1300203 VIOS_1
-------------------------------------------------------------------------
/dev/hdisk10 154.6GB 2/2 ITSO_i_CG.snap 4495 1300203 VIOS_1
_group_00001.I
TSO_i_2
-------------------------------------------------------------------------
/dev/hdisk11 154.6GB 2/2 ITSO_i_7 4359 1300203 VIOS_1
-------------------------------------------------------------------------
/dev/hdisk12 154.6GB 2/2 ITSO_i_8 4360 1300203 VIOS_1
VTD vtscsi0
Status Available
LUN 0x8100000000000000
Backing device hdisk5
Physloc
U789D.001.DQD904G-P1-C1-T1-W5001738000CB0160-L1000000000000
Mirrored false
VTD vtscsi1
Status Available
LUN 0x8200000000000000
Backing device hdisk6
Physloc
U789D.001.DQD904G-P1-C1-T1-W5001738000CB0160-L2000000000000
Mirrored false
3. For a particular virtual SCSI device, observe the corresponding LUN ID by using VIOS
command lsdev -dev vtscsix -vpd. In the example, the virtual LUN id of device
vtscsi0, is 1, as can be seen in Figure 8-13.
$
$ lsdev -dev vtscsi0 -vpd
vtscsi0 U9117.MMA.06C6DE1-V15-C20-L1 Virtual Target Device - Disk
$
4. In IBM i command STRSST to start the use system service tools (SST). You need the SST
user ID and password to sign in to the SST. After you are in SST, perform these steps:
a. Select Option 3. Work with disk units Option 1. Display disk configuration
Option 1. Display disk configuration status.
b. In the Disk Configuration Status panel, use F9 to display disk unit details.
During experimentation, the same capacity was always used. For some experiments, a few
large volumes were used. For others, a larger number of smaller volumes were used.
Specifically, a 6-TB capacity was used. The capacity was divided into 6 * 1-TB volumes, or
into 42 * 154-GB volumes.
The experimentation is intended to show the performance improvement for an IBM i workload
when running on XIV Gen 3 compared to an XIV generation 2 system. Tests with large and
small numbers of LUNs was run on both XIV Gen 3 and XIV generation 2 systems. Both
systems are equipped with 15 modules.
Remember: The purpose of the tests is to show the difference in IBM i performance
between using a few large LUNs and many smaller LUNs. They also compare IBM i
performance between XIV Gen 3 and XIV generation 2 systems.
The goal is not to make an overall configuration and setup recommendation for XIV to
handle a specific IBM i workload.
This environment is used to connect to both the XIV Storage system generation 2 and XIV
Gen 3.
Remember: In all the tests except the test with combined double workload, the XIV is
exclusively used by one IBM i LPAR. In other words, no other applications or server I/O is
running on the XIV.
In the test with combined double workload, the XIV is used only by the two IBM i LPARs.
Again, no other workloads are on the XIV.
POWER7 XIV
VIOS1
Modules
Disk pool Disk pool
LPAR3
LPAR2
VIOS2
VIOS2
Figure 8-17 shows the XIV volumes reporting in IBM i SST for the 6 * 1-TB LUNs
configuration.
..................................................................
The CPW application simulates the database server of an online transaction processing
(OLTP) environment. These transactions are all handled by batch server jobs. They represent
the type of transactions that might be done interactively in a client environment. Each of the
transactions interacts with three to eight of the nine database files that are defined for the
workload. Database functions and file sizes vary.
Functions exercised are single and multiple row retrieval, single and multiple row insert, single
row update, single row delete, journal, and commitment control. These operations are run
against files that vary from hundreds of rows to hundreds of millions of rows. Some files have
multiple indexes, whereas some have only one. Some accesses are to the actual data and
some take advantage of advanced functions such as index-only access.
After the workload is started it generates the jobs in the CPW subsystems. Each job
generates transactions. The CPW transactions are grouped by regions, warehouses, and
users. Each region represents 1000 users or 100 warehouses: Each warehouse runs 10
users. CPW generates commercial types of transactions such as orders, payments, delivery,
end stock level.
Table 8-2 shows the number of different transaction types, the percentage of each type of
transaction, the average response time, and the maximal response time. The average
response time for most of the transactions is between 0.03 and 0.04 seconds, and the
maximum response time is 11.4 seconds.
14000.0
12000.0
10000.0
8000.0 Reads/sec
6000.0 Writes/sec
4000.0
2000.0
0.0
02
02
02
02
02
02
02
02
02
02
02
:0
:0
1:
6:
6:
1:
1:
6:
6:
6:
6:
1:
6:
:01
:41
:5
:5
:3
:3
:4
:0
:1
:1
:2
:2
:4
21
21
22
22
22
22
22
22
22
22
21
22
22
154GB LUNs XIV Gen 2 - Serv Time
6.0
5.0
4.0
3.0 Serv Time
2.0
1.0
0.0
2
02
02
02
02
02
02
02
02
02
02
02
02
21 e
:0
im
1:
6:
1:
6:
1:
6:
1:
6:
6:
1:
6:
1:
:56
tT
:0
:0
:1
:1
:4
:5
:2
:2
:3
:3
:4
:4
21
22
22
22
22
22
22
22
22
22
22
21
ar
St
al
rv
te
In
During this collection period, CPW experienced an average of 8355 reads/sec and 11745
writes/sec. The average service time was 4.4 ms.
Tip: Because the reported disk wait time in IBM i collection services reports was 0 in all the
tests, it is not included in the graphs.
Figure 8-20 on page 241, Figure 8-21 on page 242 and Figure 8-22 on page 243 show the
following values reported in XIV:
I/O rate
Latency
Bandwidth in MBps
Read/write ratio
Cache hits
Tip: For readers who cannot see the colors, the various data and scales are labeled in
Figure 8-20. The other graphs are similar.
Table 8-3 shows the number of different transaction types, the percentage of each type of
transaction, the average response time, and the maximal response time. The average
response time for most of the transactions is between 0.3 and 10 seconds. The maximal
response time is 984.2 seconds.
9000.0
8000.0
7000.0
6000.0
5000.0 Reads/sec
4000.0 Writes/sec
3000.0
2000.0
1000.0
0.0
19
19
19
19
19
19
19
19
19
19
19
19
19
1:
6:
1:
6:
1:
6:
1:
1:
6:
1:
6:
6:
1:
:2
:2
:3
:4
:4
:5
:5
:0
:0
:1
:2
:3
:1
15
15
15
15
15
15
15
15
16
16
16
16
16
1 TB LUNs XIV Gen 2- Service Time (ms)
16.0
14.0
12.0
10.0
8.0 Serv Time
6.0
4.0
2.0
0.0
19
19
19
19
19
19
19
19
19
19
19
19
e
19
m
1:
1:
6:
1:
6:
1:
6:
1:
6:
1:
6:
1:
6:
Ti
:2
:3
:3
:4
:4
:5
:5
:0
:0
:1
:1
:2
:2
rt
15
15
15
15
15
15
15
15
16
16
16
16
16
ta
S
al
rv
te
In
The restore of CPW database, which was performed at the end of the run, took 24 minutes.
Figure 8-24 on page 245, Figure 8-25 on page 246 and Figure 8-26 on page 247 show the
following values reported in XIV:
I/O rate
Latency
Bandwidth in MBps
Read/write ratio
Cache hits
16000.0
14000.0
12000.0
10000.0 Reads/sec
8000.0
6000.0 Writes/sec
4000.0
2000.0
0.0
24
24
24
24
24
24
e
24
24
24
24
24
24
m
7:
7:
7:
2:
7:
2:
7:
2:
7:
2:
2:
2:
Ti
:4
:1
:2
:3
:4
:5
:5
:0
:0
:1
:2
:3
rt
12
12
13
13
13
13
13
13
13
13
12
13
ta
S
al
rv
te
In
0.7
0.6
0.5
0.4
Serv Time
0.3
0.2
0.1
0.0
24
24
24
24
24
24
24
24
24
24
24
24
e
im
7:
7:
2:
7:
2:
7:
2:
7:
2:
7:
2:
2:
tT
:4
:0
:1
:2
:3
:5
:5
:0
:1
:2
:3
:4
ar
12
13
13
13
13
13
13
13
13
12
12
13
St
al
rv
te
In
Figure 8-28 shows these values during the whole CPW run.
16000.0
14000.0
12000.0
10000.0
Reads/sec
8000.0
6000.0 Writes/sec
4000.0
2000.0
0.0
27
27
27
27
27
27
27
27
27
27
27
e
im
:2
1:
6:
1:
6:
1:
6:
1:
6:
1:
6:
1:
:46
tT
:5
:5
:1
:3
:3
:4
:0
:0
:1
:2
:2
ar
15
15
15
15
15
16
16
16
16
15
16
16
St
al
rv
te
In
0.7
0.6
0.5
0.4
Serv Time
0.3
0.2
0.1
0.0
27
27
27
27
27
27
27
27
27
27
27
27
e
im
1:
1:
6:
1:
6:
1:
6:
1:
6:
6:
1:
6:
tT
:4
:1
:2
:3
:3
:4
:5
:5
:0
:0
:1
:2
15
15
15
15
16
16
16
16
16
16
ar
15
15
St
al
rv
te
In
Figure 8-32 shows these values during the whole CPW run.
Table 8-6 shows the number of different transaction types, the percentage of each type of
transaction, the average response time, and the maximal response time. The average
response time for most of the transactions varies from 0.6 to 42 seconds. The maximal
response time is 321 seconds.
8000
7000
6000
5000 Reads/sec
4000
3000 Writes/sec
2000
1000
0
18
18
18
18
18
18
18
18
18
18
18
18
18 e
im
7:
2:
7:
7:
7:
2:
7:
2:
7:
2:
2:
2:
T
:2
:3
:4
:1
:2
:3
:4
:5
:5
:0
:0
:1
rt
18
18
18
18
19
19
19
19
18
18
19
S ta
al
rv
te
In
10.0
9.0
8.0
7.0
6.0
5.0 Serv Time
4.0
3.0
2.0
1.0
0.0
18
18
18
18
18
18
18
18
18
18
18
18
e
im
7:
7:
2:
7:
2:
7:
2:
7:
2:
7:
2:
2:
tT
:2
:4
:5
:0
:1
:3
:3
:4
:5
:0
:1
:2
ar
18
18
18
18
18
19
19
19
19
18
18
19
St
al
rv
te
In
Figure 8-35 1-TB LUNs, combined double workload, I/O rate, and service time
The restore of CPW database, which is performed at the end of the run, took 16 minutes.
Table 8-7 shows the transaction response time for the CPW run on the 154-GB LUNs.
Average response time for most of the transactions is between 1.6 and 12 seconds.
35000
30000
25000
20000 Reads/sec
15000 Wri tes /sec
10000
5000
0
9
e
18 9
18 9
18 9
:39
18 9
19 9
19 9
19 9
:39
im
:3
:3
:3
:3
5:3
:3
0:3
5:3
5:
0:
:30
:40
:45
:55
:05
:20
tT
:00
:2
:5
:1
:3
:1
ar
18
18
18
19
19
St
al
rv
te
In
4.9
4.8
4.7
4.6
Serv Ti me
4.5
4.4
4.3
4.2
9
9
e
39
18 9
18 9
:39
19 9
:39
39
19 9
39
im
:3
:3
:3
:3
:3
5:3
0:3
5:
5:
0:
:35
:40
:50
:00
:15
tT
:30
:45
:2
:5
:0
:1
:2
ar
18
18
18
18
18
19
19
19
St
al
rv
te
In
Figure 8-36 154-GB LUNs, combined double workload, I/O rate, and service time
In this test, workloads were run in both IBM i LPARs at the same time. Figure 8-37 on
page 258, Figure 8-38 on page 259 and Figure 8-39 on page 260 show the following values
in XIV:
I/O rate
Latency
Bandwidth in MBps
Cache hits
Figure 8-37 XIV values for entire run of both CPW workloads
Figure 8-38 XIV values for data collection of both CPW workloads
Figure 8-39 XIV values for database restore of both CPW workloads
XIV Gen 2
XIV Gen 3
XIV Gen 3
Concurrent
double workload
Whether using small LUNs or few large LUNs on an XIV Gen 3 system, the performance is
good in both cases. There is no significant difference between the response times in the two
environments. However when the XIV Gen 3 is stressed by running double workload in both
LPARs, large difference develop in response times between the two environments. The
advantage goes to the configuration with many small LUNs.
The better performance with many small LUNs is for the following reasons:
Queue-depth is the number of I/O operations that can be done concurrently to a volume.
The queue-depth for an IBM i volume in VIOS is a maximum of 32. This maximum is
modest comparing to the maximal queue-depths for other open servers. Therefore, in
IBM i, define a larger number of small LUNs than for other open system servers. This
configuration provides a comparable number of concurrent IO operations for the disk
space available.
The more LUNs that are available to an IBM i system, the more server tasks IBM i storage
management uses to manage the I/O operations to the disk space. Therefore, better I/O
performance is achieved.
When the increased workload of192,000 users was run concurrently in the two LPARs, cache
hits fell 60% - 80%. Disk service times increased to 4 - 8 ms.
In each configuration, the number of LUNS is a multiple of 6. For a fully configured XIV
System with six Interface Modules, this configuration equally distributes the workload (I/O
traffic) across the Interface Modules.
Important: The procedures and instructions given here are based on code that was
available at the time of writing this book. For the latest support information and instructions,
see the System Storage Interoperability Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
The IBM XIV Storage System, with its grid architecture, automated load balancing, and ease
of management, provides best-in-class virtual enterprise storage for virtual servers. It also
provides the following advantages to help meet your enterprise virtualization goals:
IBM XIV end-to-end support for VMware solutions, including vSphere and VCenter
Provides hotspot-free server-storage performance
Optimal resource use
An on-demand storage infrastructure that allows simplified growth
IBM collaborates with VMware on the strategic, functional, and engineering levels. IBM XIV
system uses this technology partnership to provide robust solutions and release them quickly.
The XIV system is installed at VMware Reference Architecture Lab and other VMware
engineering development labs. It is used for early testing of new VMware product release
features. Among other VMware product projects, IBM XIV took part in the development and
testing of VMware ESX 4.1.
IBM XIV engineering teams have ongoing access to VMware co-development programs such
as developer forums. They also have access to a comprehensive set of developer resources
including toolkits, source code, and application programming interfaces. This access
translates to excellent virtualization value for customers.
For more information, see A Perfect Fit: IBM XIV Storage System with VMware for Optimized
Storage-Server Virtualization, available at:
ftp://ftp.hddtech.ibm.com/isv/A_Perfect_Fit_IBM_XIV_and_VMware.pdf
When the VMware ESX server virtualizes a server environment, the VMware Site Recovery
Manager allows administrators to automatically fail over the whole environment or parts of it
to a backup site.
VMware Site Recovery Manager uses the mirroring capabilities of the underlying storage to
create a copy of the data at a second location. This location can be, for example, a backup
IBM XIV Storage System has a Storage Replication Adapter that integrates the IBM XIV
Storage System with VMware Site Recovery Manager.
For more information about SRM, see 9.7, XIV Storage Replication Adapter for VMware
SRM on page 318, and Appendix A, Quick guide for VMware Site Recovery Manager on
page 417.
VAAI support
ESX/ESXi 4.1 brought a new level of integration with storage systems through the use of
vStorage API for Array Integration (VAAI). VAAI helps reduce host usage, and increases
scalability and the operational performance of storage systems. The traditional ESX
operational model with storage systems forced the ESX host to issue many identical
commands to complete certain types of operations, including cloning operations. Using VAAI,
the same task can be accomplished with far fewer commands.
For more information, see 9.5, VMware vStorage API Array Integration (VAAI) on page 298.
VMware vSphere 5
VMware released vSphere 5.0 in July 2011. VMware vSphere 5.0 consists of version 5 of
ESXi and version 5 of the vCenter server. From a storage perspective, many significant
enhancements in vSphere 5.0 have made storage easier to manage:
Virtual Machine file system version 5 (VMFS-5). Enhancements provided in VMFS-5
include:
Unified 1 MiB File Block Size. There is no longer a requirement to use 1, 2, 4 or 8-MB
file blocks to create large files (files greater than 256 GB). When creating a VMFS-5
data store, the user is no longer prompted to select a maximum file size.
Large Single Extent Volumes. Before VMFS-5, the largest volume XIV could present to
an ESX server was 2 TiB. With VMFS-5, this limit is increased to 64 TiB
(64*1024*1024*1024*1024 bytes).
Small Sub-Block. VMFS-5 introduces a subblock of 8 KB rather than 64 KB. This
smaller size means smaller files use less space.
Small File Support. For files equal to or smaller than 1 KB, VMFS-5 uses the file
descriptor location in the metadata for storage rather than subblocks. When the file
grows above 1 KB, it starts to use an 8-KB subblock.
Increased File Count. VMFS-5 introduces support for more than 100,000 files
compared to around 30,000 in VMFS-3.
Atomic Test & Set (ATS) was introduced with VAAI. It is now used throughout VMFS-5
for file locking, providing improved file locking performance over previous versions of
VMFS.
Support for larger raw device mappings (RDMs). Pass through RDMs can now be 64 TiB
in size (as opposed to 2 TiB in VSphere 4.1).
Improvements to storage vMotion using a new mirror driver mechanism.
VAAI improvements including:
Removing the requirement to install a vendor supplied VAAI driver.
Adding a new primitive called Thin Provisioning Block Space Reclamation using the
T10 SCSI unmap command. This primitive allows space reclamation where thin
provisioning is used.
For more information about implementing ESXi 5.0 with XIV, see 9.4, VMware ESXi 5.0 and
XIV on page 292.
You can implement a high availability solution in your environment by adding and deploying
an additional server (or servers) running under VMware ESX. Also, implement the VMware
High Availability option for your ESX servers.
To further improve the availability of your virtualized environment, simplify business continuity
and disaster recovery solutions. Implement ESX servers, vCenter server, and another XIV
storage system at the recovery site. Also, install VMware Site Recovery Manager and use the
Storage Replication Adapter to integrate VMware Site Recovery Manager with your XIV
storage systems at both sites. The Site Recovery Manager itself can also be implemented as
a virtual machine on the ESX server.
Finally, you need redundancy at the network and SAN levels for all your sites.
Network LAN 2
Network LAN 2
VM VM VM VM VM VM VM VM VM VM VM
VM vCenter server
vCenter server SRM server
SRM server Communications between SRM servers SRA
SRA
DB for:
DB for: vCenter server
vCenter SRM server
server
SRM server
.... ....
Network LAN 1
Network LAN 1
Figure 9-1 Virtualized environment built on the VMware and IBM XIV Storage System
The rest of this chapter is divided into a number of major sections. The first three sections
address specifics for VMware ESX 3.5, ESX/ ESXi 4.x, and ESXi 5.1. The chapter also
reviews several IBM XIV integration features with VMware:
VAAI
The IBM Storage Management Console for VMware vCenter
The XIV Storage Replication Adapter for VMware Site Recovery Manager
Details about Fibre Channel configuration on VMware ESX server 3.5 can be found at:
http://www.vmware.com/pdf/vi3_35/esx_3/r35u2/vi3_35_25_u2_san_cfg.pdf
Follow these steps to configure the VMware host for FC attachment with multipathing:
1. Check host bus adapters (HBAs) and Fibre Channel (FC) connections from your host to
XIV Storage System.
2. Configure the host, volumes, and host mapping in the XIV Storage System.
3. Discover the volumes created on XIV.
Supported FC HBAs are available from IBM, Emulex, and QLogic. Further details about
HBAs supported by IBM are available from the SSIC website at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
Unless otherwise noted in the SSIC, use the firmware and driver versions promoted by
VMware in association with the relevant hardware vendor. You can find supported VMware
driver versions at:
http://www.vmware.com/resources/compatibility/search.php?deviceCategory=io
Tip: If Windows 2003 guests are using LSILogic drivers, see the following VMware
knowledge base topic regarding blocksize: http://kb.vmware.com/kb/9645697. Generally,
use a maximum block size of 1 MB.
Group ESX hosts that access the same shared LUNs in a cluster (XIV cluster) as shown in
Figure 9-2.
3. Select Scan for New Storage Devices and Scan for New VMFS Volumes, then click OK
as shown in Figure 9-5.
In this example, controller vmhba2 can see two LUNs (LUN 0 and LUN 1) circled in green.
These LUNs are visible on two targets (2 and 3) circled in red. The other controllers on the
host show the same path and LUN information.
For detailed information about how to use LUNs with virtual machines, see the VMware
guides, available at:
http://www.vmware.com/pdf/vi3_35/esx_3/r35u2/vi3_35_25_u2_admin_guide.pdf
http://www.vmware.com/pdf/vi3_35/esx_3/r35u2/vi3_35_25_u2_3_server_config.pdf
The reason for this requirement is the risk of resignature thrashing related to the LUN ID, not
the target. This restriction is described in the topic at http://kb.vmware.com/kb/1026710.
While the title of the topic refers to ESX 4.x hosts, it also addresses ESX 3.5.
By mapping volumes to the cluster rather than to each host, you ensure that each host
accesses each volume using the same LUN ID. Private mappings can be used if necessary.
VMware provides its own multipathing I/O driver for ESX. No additional drivers or software are
required. As such, the XIV Host Attachment Kit provides only documentation, and no software
installation is required.
ESX Native Multipathing automatically detects IBM XIV and sets the path policy to Fixed by
default. Do not change this setting. Also, when setting the preferred path or manually
assigning LUNs to specific path, monitor it carefully so you do not overload the IBM XIV
storage controller port. Use esxtop to monitor outstanding queues pending execution.
XIV is an active/active storage system, and therefore it can serve I/O to all LUNs using every
available path. However, the driver with ESX 3.5 cannot perform the same function and by
default cannot fully load balance. You can artificially overcome this limitation by confirming the
correct pathing policy (correcting if necessary) and distributing the I/O load over the available
HBAs and XIV ports. This process is called as manual load balancing. To manually balance
the load, follow these instructions:
1. Set the pathing policy in ESX 3.5 to either Most Recently Used (MRU) or Fixed. When
accessing storage on the XIV, the correct policy is Fixed. In the VMware Infrastructure
Client, select the server, click the Configuration tab, and select Storage (Figure 9-7).
In this example, the LUN is highlighted (esx_datastore_1) and the number of paths is 4
(circled in red).
3. To change the current path, select Manage Paths and the Manage Paths window opens
as shown in Figure 9-9. Set the pathing policy to Fixed if it is not already by selecting
Change in the Policy window.
5. Repeat steps 1-4 to manually balance I/O across the HBAs and XIV target ports. Because
workloads change over time, you will need reviewed the balance periodically.
Example 9-1 and Example 9-2 show the results of manually configuring two LUNs on
separate preferred paths on two ESX hosts. Only two LUNs are shown for clarity, but this
configuration can be applied to all LUNs assigned to the hosts in the ESX datacenter.
Example 9-2 shows the results of manually configuring two LUNs on separate preferred paths
on ESX host 2.
As an alternative to manually setting up paths to load balance, contact your IBM Technical
Advisor or pre-sales technical support for a utility called xivcfg-fixedpath. This utility can be
used to achieve an artificial load balance of XIV LUNs on VMware ESX 3.X and later.
The utility uses standard esxcfg and esxcli commands, and is run like a script as shown in
Example 9-3. This utility is not available for download through the internet.
# ./xivcfg-fixedpath -V
xivcfg-fixedpath: version 1.2
#./xivcfg-fixedpath
----------------------------------------------------------------------------------
This program will rescan all FC HBAs, change all XIV disks to Fixed path policy
and reassign the XIV preferred disk path to balance all XIV LUNs across available
paths. This may result in I/O interruption depending on the I/O load and state of
devices and paths. Proceed (y/n)?
Supported FC HBAs are available from Brocade, Emulex, IBM, and QLogic. Further details
about HBAs supported by IBM are available from the SSIC website:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
Tip: If Windows 2003 guests are using LSILogic drivers, see the following VMware
knowledge base topic regarding block size:
http://kb.vmware.com/kb/9645697
Repeat this process for all ESX hosts that you plan to connect to the XIV Storage System.
After identifying the ESX host port WWNs, you are ready to define hosts and clusters for the
ESX servers. Create LUNs and map them to defined ESX clusters and hosts on the XIV
Storage System. See Figure 9-2 on page 269 and Figure 9-3 on page 269 for how this
configuration might typically be set up.
Tip: Group the ESX hosts that access the same LUNs in a cluster (XIV cluster) and the
assign the LUNs to that cluster.
If each XIV volume can be accessed through 12 fabric paths (which is a large number of
paths), then the maximum number of volumes is 85. Dropping the path count to six increases
the maximum LUN count to 170. For installations with large numbers of raw device mappings,
these limits can become a major constraint.
3. The new LUNs are displayed in the Details pane as depicted in Figure 9-16.
In the example, t controller vmhba1 can see two LUNs (LUN 1 and LUN 2) circled in red. The
other controllers on the host show the same path and LUN information.
The procedures and instructions given here are based on code that was available at the time
of writing. For the latest support information, see the Storage System Interoperability Center
(SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
With the release of VMware ESX 4 and VMware ESXi 4, VMware introduced the concept of a
Pluggable Storage Architecture (PSA). PSA in turn introduced additional concepts to its
Native Multipathing Plugin (NMP). The PSA interacts at the VMkernel level. It is an open and
modular framework that can coordinate the simultaneous operations across multipathing
solutions.
VMware NMP chooses the multipathing algorithm based on the storage system type. The
NMP associates a set of physical paths with a specific storage device or LUN. The NMP
module works with sub plug-ins such as a Path Selection Plug-In (PSP) and a Storage Array
Type Plug-In (SATP). The SATP plug-ins are responsible for handling path failover for a
storage system. The PSPs plug-ins are responsible for determining which physical path is
used to issue an I/O request to a storage device.
Note: Starting with XIV software Version 10.1, the XIV Storage System is a T10 ALUA-
compliant storage system.
ESX/ESXi 4.x automatically selects the appropriate SATP plug-in for the IBM XIV Storage
System based on the XIV Storage System software version. For versions before 10.1 and for
ESX 4.0, the Storage Array Type is VMW_SATP_DEFAULT_AA. For XIV versions later than
10.1 and with ESX/ESXi 4.1, the Storage Array Type is VMW_SATP_DEFAULT_ALUA.
PSPs run with the VMware NMP, and are responsible for choosing a physical path for I/O
requests. The VMware NMP assigns a default PSP to each logical device based on the SATP
associated with the physical paths for that device.
ESX has built-in rules defining relations between SATP and PSP for the storage system.
VMKernel
Vmware PSP
VMware SATP (ALUA)
(Most Recently Used)
Before proceeding with the multipathing configuration, complete the tasks described in the
following sections:
9.3.1, Installing HBA drivers on page 276
9.3.2, Identifying ESX host port WWN on page 277
Considerations for the size and quantity of volumes on page 277.
Here you can see data store currently defined for the ESX host.
4. Click Add Storage to open the window shown in Figure 9-19.
6. Select the LUN that you want to use as a new data store, then click Next. A new window
like Figure 9-21 is displayed.
Figure 9-21 shows the partition parameters that are used to create the partition. If you
need to change the parameters, click Back. Otherwise, click Next.
7. The window shown in Figure 9-22 displays. Type a name for the new data store and click
Next. In this example, the name is XIV_demo_store.
Figure 9-23 Selecting the file system parameters for ESX data store
Tip: For more information about selecting the right values for file system parameters for
your specific environment, see your VMWare ESX/ESXi 4.x documentation.
9. In the summary window shown in Figure 9-24, check the parameters that you entered. If
everything is correct, click Finish.
Set up the round-robin policy for the new data store by following these steps:
1. From the vSphere Client main window (Figure 9-26), you can see a list of all data store,
including the new one you created. Select the data store you want to change the policy on,
then click Properties
3. The Manage Paths window shown in Figure 9-28 is displayed. Select any of the vmhbas
listed.
5. Click Change to confirm your selection and return to the Manage Paths window as shown
in Figure 9-30.
Figure 9-30 edata store paths with selected round robin policy for multipathing
Apply the round-robin policy to any previously created data stores. Your ESX host is now
connected to the XIV Storage System with the correct settings for multipathing.
Tip: Commands using esxcli need either the vSphere CLI installed on a management
workstation or the Tech Support Mode enabled on the ESX server itself. Enabling the Tech
Support Mode allows remote SSH shell access. If esxcli is run from a command prompt
without any form of configuration file, each command normally uses the following syntax:
esxcli --server 9.155.113.135 --username root --password passw0rd <command>
If you run esxcli from a Tech Support Mode shell or on a host with UNIX utilities, you can
use commands like grep and egrep. For more information, see the following knowledge
base topics:
http://kb.vmware.com/kb/1003677
http://kb.vmware.com/kb/1017910
http://kb.vmware.com/kb/2004746
For Qlogic HBAs, verify which Qlogic HBA module is currently loaded as shown in
Example 9-5.
3. Set the new value for the queue_depth parameter and check that new values are applied.
For Emulex HBAs, see Example 9-6.
Example 9-6 Setting new value for queue_depth parameter on Emulex FC HBA
# esxcfg-module -s lpfc0_lun_queue_depth=64 lpfc820
# esxcfg-module -g lpfc820
lpfc820 enabled = 1 options = 'lpfc0_lun_queue_depth=64
Example 9-7 Setting new value for queue_depth parameter on Qlogic FC HBA
# esxcfg-module -s ql2xmaxqdepth=64 qla2xxx
# esxcfg-module -g qla2xxx
qla2xxx enabled = 1 options = 'ql2xmaxqdepth=64
You can also change the queue_depth parameters on your HBA using the tools or utilities
provided by your HBA vendor.
Important: The default ESX VMware settings for round-robin are adequate for most
workloads. Do not change them normally.
If you need to change the default settings, enable the non-optimal use for round-robin and
decrease the amount of I/O going over each path. This configuration can help the ESX host
use more resources on the XIV Storage System.
Here you can view the device identifier for your data store (circled in red).
4. Log on to the service console as root or access the esxcli. You can also get the device IDs
using the esxcli as shown in Example 9-8.
5. Enable use of non-optimal paths for round-robin with the esxcli command as shown in
Example 9-9.
Example 9-9 Enabling use of non-optimal paths for round-robin on ESX/ESXi 4.x host
#esxcli nmp roundrobin setconfig --device eui.0017380000691cb1 --useANO=1
6. Change the amount of I/O run over each path as shown in Example 9-10. This example
uses a value of 10 for a heavy workload. Leave the default (1000) for normal workloads.
Example 9-10 Changing the amount of I/O run over one path for round-robin algorithm
# esxcli nmp roundrobin setconfig --device eui.0017380000691cb1 --iops=10
--type "iops"
Example 9-12 Setting round-robin tweaks for all IBM XIV Storage System devices
# script to display round robin settings
for i in `ls /vmfs/devices/disks/ | grep eui.001738*|grep -v \:` ; \
do echo "Current round robin settings for device" $i ; \
esxcli nmp roundrobin getconfig --device $i
done
5. If you use VMFS-5, you are not prompted to define a maximum block size. You are given
the option to use a custom space setting, limiting the size of the data store on the volume.
You can expand the data store at a later time to use the remaining space on the volume.
However, you cannot use that space for a different data store.
There is no need to manage the paths to the XIV because round robin should already be
in use by default.
These facts have some important design considerations. If each XIV volume can be accessed
through 12 fabric paths (which is a large number of paths), the maximum number of volumes
is 85. Dropping the paths to a more reasonable count of six increases the maximum LUN
count to 170. For installations with large numbers of raw device mappings, these limits can
become a major constraint.
If esxcli is run from a Tech Support Mode shell or on a host with UNIX utilities, commands
like grep and egrep can be used. For more information, see:
http://kb.vmware.com/kb/1017910
http://kb.vmware.com/kb/2004746
2. Set the queue depth for the relevant HBA type. In both Example 9-14 and Example 9-15,
the queue depth is changed to 64. In Example 9-14 the queue depth is set for an Emulex
HBA.
Example 9-14 Setting new value for queue_depth parameter on Emulex FC HBA
# esxcli system module parameters set -p lpfc0_lun_queue_depth=64 lpfc820
Example 9-15 Setting new value for queue_depth parameter on Qlogic FC HBA
# esxcli system module parameters set -p ql2xmaxqdepth=64 -m qla2xxx
3. Reboot your ESXi server. After reboot, confirm that the new settings are applied using the
command shown in Example 9-16. The example shows only one of a great many
parameters. You need to change the syntax if you have an Emulex HBA.
Example 9-16 Checking the queue depth setting for a Qlogic HBA
# esxcli system module parameters list -m qla2xxx | grep qdepth
Name Type Value Description
-------------------------- ---- ----- -----------
ql2xmaxqdepth int 64 Maximum queue depth
Important: The default ESX VMware settings for round-robin are adequate for most
workloads and should not normally be changed.
If you need to change the default settings, enable the non-optimal use for round-robin and
decrease the amount of I/O going over each path. This configuration can help the ESX host
use more resources on the XIV Storage System.
3. Change the amount of I/O run over each path as shown in Example 9-18. This example
uses a value of 10 for a heavy workload. Leave the default (1000) for normal workloads.
Example 9-18 Changing the amount of I/O run over one path for round-robin algorithm
# esxcli storage nmp psp roundrobin deviceconfig set --iops=10 --type "iops"
--device eui.0017380000cb11a1
If you need to apply the same settings to multiple data stores, you can also use scripts similar
to the ones shown in Example 9-20.
Example 9-20 Setting round-robin tweaks for all IBM XIV Storage System devices
# script to display round robin settings
for i in `ls /vmfs/devices/disks/ | grep eui.001738*|grep -v \:` ; \
do echo "*** Current settings for device" $i ; \
esxcli storage nmp device list --device $i
done
Do not create a single giant data store rather than multiple smaller ones for the following
reasons:
Each XIV volume is assigned a SCSI queue depth by ESXi. More volumes mean more
SCSI queues, which means more commands can be issued at any one time.
The maximum number of concurrent storage vMotions per data store is still limited to 8.
If you want to create a data store that approaches the maximum size, limit the maximum XIV
volume size as follows:
2nd generation XIV 70368 GB (65536 GiB or 137438953472 Blocks)
XIV Gen 3 70364 GB (65531 GiB or 137428467712 Blocks)
Example 9-21 shows the largest possible data store, which is exactly 64 TiB in size. The df
command was run on the ESX server using the tech support mode shell.
IBM XIV with the correct firmware release supports the following T10-compliant SCSI
commands (also called primitives) to achieve this new level of integration:
Hardware Accelerated Move: This command offloads copy operations from VMware ESX
to the IBM XIV Storage System. This process allows for rapid movement of data when
performing copy, move, and VMware snapshot operations within the IBM XIV Storage
System. It reduces the processor and HBA workload of the ESX server. Similarly, it
reduces the volume of traffic moving through the SAN when performing VM deployment. It
does so by VM cloning and storage cloning at the block level within and across LUNs. This
command has the following benefits:
Expedited copy operations
Minimized host processing/resource allocation
Reduced network traffic
A considerable reduction in elapsed time to perform these tasks
This command works only when the source and target LUNs are on the same XIV.
Hardware accelerated initialization: This command reduces server processor and HBA
workload. It also reduces the volume of SAN traffic when performing repetitive block-level
write operations within virtual machine disks to IBM XIV. Block Zeroing allows the VMware
host to save bandwidth and communicate faster by minimizing the amount of actual data
sent over the path to IBM XIV. Similarly, it allows IBM XIV to minimize its own internal
bandwidth consumption and virtually apply the write much faster.
Hardware Assisted Locking (also known as Atomic Test & Set or ATS): Intelligently
relegates resource reservation down to the selected block level to lock an entire LUN
during VMware metadata updates. It does this instead of using an SCSI2 reserve. This
process has the following advantages. These advantages are obvious in enterprise
environments where LUNs are used by multiple applications or processes at once.
Significantly reduces SCSI reservation contentions
Lowers storage resource latency
Enables parallel storage processing
With ESX/ESXi 4.1, you must install an IBM supplied plugin on each vSphere server. The
initial release was version 1.1.0.1, which supported the XIV. IBM then released version
1.2.0.0, which added support for Storwize V7000 and SAN Volume Controller.
The IBM Storage Device Driver for VMware, the release notes, and the Installation Guide can
be downloaded at:
http://www-933.ibm.com/support/fixcentral/swg/selectFixes?parent=ibm/Storage_Disk&
product=ibm/Storage_Disk/XIV+Storage+System+(2810,+2812)&release=All&platform=All&
function=all
With vSphere 5.0, you do not need to install a vendor supplied driver to enable VAAI. It is
supported natively.
9.5.2 Installing the IBM VAAI device driver on an ESXi 4.1 server
The IBM Storage device driver for VMware VAAI is a kernel module that allows the VMware
VAAI driver to offload certain storage operations to the storage hardware. In this example,
they are offloaded to an XIV. The driver needs to be installed on every ESX/ESXi 4.1 server
and requires that each server is restarted after installation. Updates to the IBM Storage driver
also require that each ESX/ESXi server is rebooted. When combined with server vMotion and
vSphere server redundancy, this process usually does not require any guest host outages.
IBM has so far released two versions of the driver that are named as follows:
Version 1.1.0.1 IBM-ibm_vaaip_module-268846-offline_bundle-395553.zip
Version 1.2.0.0 IBM-ibm_vaaip_module-268846-offline_bundle-406056.zip
To confirm if the driver is already installed, use the vihostupdate.pl command with the
-query parameter as shown in Example 9-22. In this example, a version of the driver is
already installed. Because only the first 25 characters of the name are shown, it is not clear if
it is 1.1.0.1 or 1.2.0.0. If performing this command in the Tech Support Mode shell, use the
esxupdate query command.
If the driver is already installed and you have downloaded the latest version, use the -scan
-bundle command against the downloaded compressed file. This procedure checks whether
you have an older version of the driver. In Example 9-23, the bundle is not installed, indicating
that either no driver is installed, or only the older version of the driver is installed.
Example 9-23 Checking if the driver is installed or is not at the latest level
vihostupdate.pl --server 9.155.113.136 --username root --password password -scan -bundle
IBM-ibm_vaaip_module-268846-offline_bundle-406056.zip
The bulletins which apply to but are not yet installed on this ESX host are listed.
To perform the upgrade or install the driver for the first time, use server vMotion to move all
guest operating systems off the server you are upgrading. Install the new driver, place the
server in maintenance mode, and reboot it as shown in Example 9-24.
Example 9-24 Installing and then rebooting after installing the new VAAI driver
vihostupdate.pl --server 9.155.113.136 --username root --password password --install -bundle
IBM-ibm_vaaip_module-268846-offline_bundle-406056.zip
When the server reboots, confirm the driver is installed by issuing the -query command as
shown in Example 9-22 on page 299. ESXi 4.1 does not have any requirement to claim the
storage for VAAI (unlike ESX). More details about claiming IBM storage systems in ESX can
be found in the IBM VAAI driver installation guide.
All three options are enabled by default, meaning that the value of each parameter is set to 1.
If using the service console to control VAAI, the following commands were tested and found to
work on both ESX/ESXi 4.1 and ESXi 5.0. The first three commands display the status of
VAAI. If the value returned for each function is 0, that function is disabled. If the value
returned is 1, the function is enabled.
esxcfg-advcfg -g /DataMover/HardwareAcceleratedMove
esxcfg-advcfg -g /DataMover/HardwareAcceleratedInit
esxcfg-advcfg -g /VMFS3/HardwareAcceleratedLocking
The following commands disable each VAAI function (changing each value to 0):
esxcfg-advcfg -s 0 /DataMover/HardwareAcceleratedMove
esxcfg-advcfg -s 0 /DataMover/HardwareAcceleratedInit
esxcfg-advcfg -s 0 /VMFS3/HardwareAcceleratedLocking
9.5.4 Disabling and enabling VAAI on the XIV on a per volume basis
You can disable or enable VAAI support at the XIV on a per volume basis, although doing so
is normally not necessary. The commands are documented here to so that you are aware of
how it is done. Generally, do not use these commands unless advised to do so by IBM
support.
VAAI management is done using the XCLI. The two relevant commands are vol_enable_vaai
and vol_disable_vaai. If you run these commands without specifying a volume, the
command works on all volumes. In Example 9-29, VAAI is disabled for all volumes without
confirmation prompt using the -y parameter. VAAI is then enabled for all volumes, again
without confirmation.
Example 9-30 shows displaying the VAAI status for an individual volume, disabling VAAI,
confirming it is disabled, and then enabling it. The vol_list command does not show VAAI
status by default. Use the -x parameter to get the XML output. Because the XML output is
long and detailed, only a subset of the output is shown.
After you enable VAAI for your volume, you need to prompt vSphere to attempt an offload
function before hardware acceleration will show as supported. For more information, see
9.5.3, Confirming VAAI Hardware Acceleration has been detected on page 300.
Figure 9-39 Creating a virtual machine with eager zero thick without VAAI enabled
Figure 9-40 shows a virtual machine with HardwareAcceleratedInit enabled. In this test over
2200 IOPS with 1 MBps of throughput are seen for less than 30 seconds. VAAI reduced the
execution time by more than 50% and eliminated nearly all the throughput.
Figure 9-40 Creating a virtual machine with eager zero thick with VAAI enabled
Both of these tests were done with server and SAN switch hardware that was less than ideal
and no other performance tuning. These results are therefore indicative rather than a
benchmark. If your tests do not show any improvement when using VAAI, confirm that
Hardware Acceleration shows as Supported. For more information, see 9.5.3, Confirming
VAAI Hardware Acceleration has been detected on page 300.
The plug-in runs as a Microsoft Windows Server service on the vCenter server. Any VMware
vSphere Client that connects to the vCenter server detects the service on the server. The
service then automatically enables the IBM storage management features on the vSphere
Client.
9.6.1 Installation
Install the IBM Storage Management Console for VMware vCenter onto the Windows server
that is running VMWare vCenter version 4.0, 4.1 or 5.0. There are separate installation
packages for x86 and x64. You can save time by downloading the correct package for your
server architecture. During package installation, you are prompted to do the following steps:
1. Confirm which language you want to use.
2. Accept license agreements.
3. Select an installation directory location (a default location is offered).
4. When installation completes, a command-line configuration wizard starts as shown in
Example 9-31. Normally you can safely accept each prompt in the wizard. When prompted
for a user name and password, you need a user ID that is able to log on to the VMware
vSphere Client. If you do not either change or accept the SSL certificate, you get
Certificate Warning windows when starting the Client. For more information about how to
replace the SSL certificate, see the plug-in user guide.
Example 9-31 IBM Storage Management Console for VMWare vCenter Configuration wizard
Welcome to the IBM Storage Management Console for VMware vCenter setup wizard, version 2.6.0.
Use this wizard to configure the IBM Storage Management Console for VMware vCenter.
Press [ENTER] to proceed.
-------------------------------------------------------------------------------
The Wizard will now install the Management Console service and register the extension in the
vCenter server.
Do you want to continue? [default: yes ]:
-------------------------------------------------------------------------------
The IBM Storage Management Console for VMware vCenter requires a valid username for connecting
to the vCenter server.
This user should have permission to register the Plug-in in the Plug-ins Manager.
Please enter a username : Administrator
-------------------------------------------------------------------------------
Please enter the password for user Administrator :
-------------------------------------------------------------------------------
The IBM Storage Management Console for VMware vCenter web component requires a valid network
port number.
Please enter a port number for the web component [default: 8880 ]:
-------------------------------------------------------------------------------
Please wait while configuring the service and registering the extension
-------------------------------------------------------------------------------
The IBM Storage Management Console for VMware vCenter is now configured.
This product is using an SSL certificate which is not signed.
Please consult the User Guide for SSL certificate replacement instructions.
Press [ENTER] to exit.
5. After you configure the IBM Storage Management Console for VMware vCenter, restart
the client if you were already logged on. Anew IBM Storage icon plus a new IBM Storage
tab with all their associated functions are added to the VMware vSphere Client.
Maximum
Volume
Size
After you make the required registry modifications, perform these steps:
1. Close the Registry Editor.
2. Close the vSphere Client application. Users connected remotely must also close their
client applications.
3. Click Start Run to open the Windows Services window.
4. The Run dialog box is displayed. Type services.msc and then select OK.
5. Stop and then start the following Windows service IBM Storage Management Console for
VMware vCenter, as shown in Figure 9-45. You can then close the services console.
Two possible changes you might consider are in the following sections.
4. If the vSphere Client connects to the XIV successfully you get a list of storage pools on
that XIV. Select the pools that the VMware administrator will allocate storage from as
shown in Figure 9-50. Additional pools can be added later if required.
6. Select an XIV from the Storage Systems box, then select a pool from the Storage Pools
box.
7. Select the New Volume option to create a volume.
Tip: When creating a volume, use the same name for both the volume and the data
store. Using the same name ensures that the data store and volume names are
consistent.
8. You are prompted to map the volume to either individual VMware servers or the whole
cluster. Normally you would select the entire cluster.
For the plug-in to work successfully, the SAN zoning to allow SAN communication between
the VMWare cluster and the XIV must have already been completed. On the XIV, the Cluster
and hosts definitions (representing the VMware cluster and its servers) must have also been
created. This process cannot be done from the plug-in, and is not done automatically. If the
zoning and host definitions are not done, volume creation fails and the requested volume is
created and then deleted.
Tip: If you perform changes such as renaming or resizing volumes, updates might take up
to 60 seconds to display in the plug-in.
Detailed view of
snaps and mirrors
Snapshot status
Mirroring status
5. When you are prompted to enter a data store name in the Properties tab, use the same
name you used when creating the volume.
The plug-in confirms that the permission level is Read Only as shown in Figure 9-55
You are not prompted to select pools as shown in Figure 9-50 on page 312 because the user
has no authority to work with pools. However you still get to view the IBM Storage tab as
shown in Figure 9-53 on page 315. The advantage is that the VMware administrator can now
be sure which hardware matches which data store. This system allows you to identify the
following data without any ability to change or configure the XIV:
Exact XIV name
XIV serial number
XIV pool name
XIV volume name
XIV snapshot name
9.6.8 Troubleshooting
Two issues were seen during testing for the book.
Plug-in disabled
If you do not see the IBM Storage icon, you might find the plug-in is disabled. The plug-in on a
remote vSphere Client can be disabled if the client does not resolve the host name of the
vCenter server. Click Plug-ins Manage Plug-ins. If ibm-vcplugin is disabled, confirm if the
issue is that the remote name could not be resolved as shown in Figure 9-56 on page 317.
As a simple test, ping the host name listed in the error. If the remote name is the problem,
correct the issue with name resolution. One simple solution is to add an entry to the HOSTS
file of the server on which you are trying to run the vSphere Client. This issue did not occur if
To correct this error, add the relevant device using the process documented in9.6.3, Adding
IBM Storage to the plug-in on page 311. Figure 9-57 shows that the undefined system is an
XIV as indicated in the Model column. The hint as to which XIV is given in the identifier
column where we can see the identifier is: eui.00173800279502fb. This number derives from
the WWNN. The WWNN in this case would be 50:01:73:80:27:95:00:00 (note that 002795 is
the unique portion of the WWNN).
You can also determine the XIV serial by using the identifier. The first part of the identifier is
001738, which is the IEEE Object ID for IBM. The next part is 002795, which is the serial
number of the XIV in hexadecimal. If you convert that number from hex to decimal, you get
the serial number of the XIV. In this example, it is 10133.
Mirroring functions of the storage subsystems create a copy of the data on the storage device
at the backup location. In a failover case, all VMs shut down at the primary site if still
possible/required. They are restarted on the ESX hosts at the backup datacenter, accessing
the data on the backup storage system. This process involves these steps:
Stopping any running VMs at the primary side
Stopping the mirroring
Making the copy accessible to the backup ESX servers
Registering and restarting the VMs on the backup ESX servers
VMware SRM can automatically perform all these steps and failover complete virtual
environments in just one click. This process saves time, eliminates user errors, and provides
a detailed documentation of the disaster recovery plan. SRM can also perform a test of the
failover plan. It can create an additional copy of the data at the backup site and start the
virtual machines from this copy without actually connecting them to any network. This feature
allows you to test recovery plans without interfering with the production environment.
A minimum setup for SRM contains two ESX servers (one at each site), a vCenter for each
site, and two storage systems (one at each site). The storage systems need to be in a copy
services relationship. Ethernet connectivity between the two datacenters is also required for
the SRM to work.
Details about installing, configuring, and using VMware Site Recovery Manager can be found
on the VMware website at:
http://www.vmware.com/support/pubs/srm_pubs.html
Integration with storage systems requires a Storage Replication Adapter specific to the
storage system. A Service Replication Adapter is available for the IBM XIV Storage System.
At the time of writing this book, the IBM XIV Storage Replication Adapter supports the
following versions of VMware SRM server:
1.0
1.0 U1
4.0 and 4.1
The storage of the data is an important aspect of the overall virtualization strategy. You must
select an appropriate storage system that provides a complete, complementary virtualized
infrastructure. The IBM XIV Storage System, with its grid architecture, automated load
balancing, and exceptional ease of management, provides best-in-class virtual enterprise
storage for virtual servers. IBM XIV and Citrix XenServer together can provide hot-spot-free
server-storage performance with optimal resources usage. Together, they provide excellent
consolidation, with performance, resiliency, and usability features that can help you reach
your virtual infrastructure goals.
XenMotion (Live migration): Allows the live migration of running virtual machines from one
physical server to another with zero downtime, continuous service availability, and
complete transaction integrity.
VMs disk snapshots: Snapshots provide a point in time image of disk state and are
useful for virtual machine backup.
XenCenter management: Citrix XenCenter offers monitoring, management, and general
administrative functions for VM from a single interface. This interface allows easy
management of hundreds of virtual machines.
Distributed management architecture: This architecture prevents a singe point of failure
from bring down all servers across an entire data center.
Conversion Tools (Citrix XenConverter): XenConverter can convert a server or desktop
workload to a XenServer virtual machine. It also allows migration of physical and virtual
servers (P2V and V2V).
High availability: This feature allows you to restart virtual machine after it was affected by a
server failure. The auto-restart functionality protects all virtualized applications and
increases the availability of business operations.
Dynamic memory control: Can change the amount of host physical memory assigned to
any running virtual machine without rebooting it. You can also start additional virtual
Windows
Windows Server 2008 64-bit & 32-bit & R2
Windows Server 2003 32-bit SP0, SP1, SP2, R2; 64-bit SP2
Windows Small Business Server 2003 32-bit SP0, SP1, SP2, R2
Windows XP 32-bit SP2, SP3
Windows 2000 32-bit SP4
Windows Vista 32-bit SP1
Windows 7
Linux
Red Hat Enterprise Linux 32-bit 3.5-3.7, 4.1-4.5, 4.7 5.0-5.3; 64-bit 5.0-5.4
Novell SUSE Linux Enterprise Server 32-bit 9 SP2-SP4; 10 SP1; 64-bit 10 SP1-SP3,
SLES 11 (32/64)
CentOS 32-bit 4.1-4.5, 5.0-5.3; 64-bit 5.0-5.4
Oracle Enterprise Linux 64-bit & 32-bit 5.0-5.4
Debian Lenny (5.0)
10.2.1 Prerequisites
To successfully attach an XenServer host to XIV and assign storage, a number of
prerequisites must be met. The following is a generic list. Your environment might have
additional requirements.
Complete the cabling.
Configure the SAN zoning.
Install any service packs and updates if required.
Create volumes to be assigned to the host.
Supported Hardware
Information about the supported hardware for XenServer can be found in the XenServer
hardware compatibility list at:
http://hcl.xensource.com/BrowsableStorageList.aspx
Enabling multipathing has different steps depending on whether the Storage Repositories
(SRs) are on the XenServer or on attached hosts.
The XenServer hosts that need to access the same shared LUNs must be grouped in a
cluster (XIV cluster), and the LUNs assigned to that cluster. Figure 10-5 shows how the
cluster is typically set up.
Creating an SR
To create an SR, follow these instructions:
1. After the host definition and LUN mappings are completed in the XIV Storage System,
open XenCenter and select a pool or host to attach to the new SR. As shown in
Figure 10-7, you can click either button highlighted with a red rectangle to create an SR.
4. Select the LUN and click Finish to complete the configuration. XenServer starts to attach
the SR, creating Physical Block Devices (PBDs) and plugging PBDs to each host in the
pool.
SONAS is configured as a multi-system gateway device built from standard components and
IBM software. The SONAS Gateway configurations are shipped in pre-wired racks made up
of internal switching components along with interface, management, and storage nodes.
IBM SONAS Gateways can now be attached to IBM XIV Storage Systems with Fibre Channel
(FC) switches or direct connected FC. The advantages of this combination are ease of use,
reliability, performance, and lower total cost of ownership (TCO).
Figure 11-1 is a schematic view of the SONAS gateway and its components, attached to two
XIV systems.
SONAS
1 2 2 4 36 48
Cloud IP Network
2 4 2
M g mt D D 1 4 2 6 38
Rs t A A 1 1 2 3 35 47
1 3
S D D 2 4 6 8 1 0 1 2 14 16 18 20 2 2 24 26 28 30 32 34 36 3 8 40 4 2 44 46 48
A A 1 3 5 7 9 11 1 3 1 5 1 7 19 2 1 2 3 2 5 2 7 2 9 3 1 33 35 3 7 3 9 41 4 3 45 4 7
1 2 2 4 36 48
2 4 2
M g mt D D 1 4 2 6 38
Rs t A A 1 1 2 3 35 47
1 3
S D D 2 4 6 8 1 0 1 2 14 16 18 20 2 2 24 26 28 30 32 34 36 3 8 40 4 2 44 46 48
A A 1 3 5 7 9 11 1 3 1 5 1 7 19 2 1 2 3 2 5 2 7 2 9 3 1 33 35 3 7 3 9 41 4 3 45 4 7
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Sy st emx 3650 M 3
1 2
3 4
Client
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Sy st emx 3650 M 3
1 2
3 4
0 1
Interface Nodes
2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 2
Sy st emx 3650 M 3
3 4
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Sy st emx 3650 M 3
1 2
3 4
Client
2005 B32
Internal Network
NK
L
S PD
0 1 2 3 8 9 10 11 1 6 1 7 1 8 19 24 2 5 2 6 2 7
4 5 6 7 1 2 13 14 15 2 0 2 1 2 2 23 28 2 9 3 0 3 1
2005 B32
NK
L
S PD
0 1 2 3 8 9 10 11 1 6 1 7 1 8 19 24 2 5 2 6 2 7
4 5 6 7 1 2 13 14 15 2 0 2 1 2 2 23 28 2 9 3 0 3 1
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 2
3 4
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 2
3 4
Storage Nodes
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Sy st emx 3650 M 3
1 2
3 4
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Sy st emx 3650 M 3
1 2
3 4
Figure 11-1 IBM SONAS with two IBM XIV Storage Systems
The SSONAS Gateway is built on several components that are connected with InfiniBand:
The management node handles all internal management and code load functions.
Interface nodes are the connection to the customer network. They provide file shares from
the solution. They can be expanded as needed for scalability.
Storage nodes are the components that deliver the General Parallel File System (GPFS)
to the interface nodes. They provide fiber connection to the IBM XIV Storage System.
When more storage is required, more IBM XIV Storage Systems can be added.
When attaching an IBM SONAS Gateway to an IBM XIV Storage System, perform the
following checks and preparations. These preparations must be done in conjunction with the
connectivity guidelines presented in Chapter 1, Host connectivity on page 1.
Check for supported versions and prerequisite
Perform and verify the cabling
Define zoning
Create Regular Storage Pool for SONAS volumes
Create XIV volumes
Create SONAS Cluster
Add SONAS Storage Nodes to cluster
Add Storage Node Fibre Channel port (WWPN) to nodes in the cluster
Map XIV volumes to the cluster
XIV SONAS
Patch Panel
Patch Panel
Storage Node 2
Storage Node 1
Restriction: Each directly connected XIV requires its own dedicated pair of storage nodes
(known as a storage pod).
If a single IBM XIV Storage System is being connected, each switch fabric must have six
available ports for Fibre Channel cable attachment to XIV. One is needed for each interface
module in XIV. Typically, XIV interface module port 1 is used for switch fabric 1, and port 3 for
switch fabric 2, as depicted in Figure 11-3.
If two IBM XIV Storage Systems are to be connected, each switch fabric must have 12
available ports for attachment to the XIV. You need six ports for each XIV.
A maximum of two XIVs can be connected to a SONAS storage node pair (known as a
storage pod).
XIV SONAS
Patch Panel
Patch Panel
Storage Node 2
Storage Node 1
Important: When connecting to a Generation 2 XIV through an 8-Gbit fabric, set the
SONAS ports to 4 Gbit to match the XIV port speed.
Zone each HBA port from the IBM SONAS Gateway Storage Nodes to all six XIV interface
modules. If you have two XIV systems, zone to all XIV interface modules as shown in
Table 11-1 and Table 11-2 on page 337. These configurations provide the maximum number
of available paths to the XIV.
An IBM SONAS gateway connected to XIV uses multipathing with round-robin feature
enabled. Round-robin means that all I/O to the XIV are spread over all available paths.
Table 11-1 shows the zoning definitions for each SSONAS Storage Node port (initiator) to all
XIV interface modules (targets) on one switch/fabric. The shaded area shows the XIV ports
and extra zones required for a second XIV attachment.
The zoning is defined such that each SONAS Storage Node has six possible paths to an
individual XIV.
Table 11-1 Zoning for switch fabric A from SONAS to XIV Storage
Port Connected To Zone Zone Zone Zone Zone Zone Zone Zone
1 2 3 4 5 6 7 8
Table 11-2 Zoning for switch fabric B from SONAS to XIV Storage
Port Connected To Zone Zone Zone Zone Zone Zone Zone Zone
1 2 3 4 5 6 7 8
Zoning is also described in the IBM Scale Out Network Attached Storage - Installation Guide
for iRPQ 8S1101: Attaching IBM SONAS to XIV, GA32-0797, which is available at:
http://publib.boulder.ibm.com/infocenter/sonasic/sonas1ic/topic/com.ibm.sonas.doc/
xiv_installation_guide.pdf
Restriction: You must use Regular Storage Pool. Thin provisioning is not supported
when attaching to the IBM SONAS Gateway.
2. Create volumes for the IBM SONAS Gateway in the storage pool. Four 4 TB volumes are
created from the Volumes window as shown in Figure 11-5 on page 339.
Remember: The volumes are 4002 GB because the XIV Storage System uses 17-GB
capacity increments.
3. Define a cluster in XIV for the SONAS Gateway so multiple SSONAS Storage Nodes can
see the same volumes. See Figure 11-7.
Create another host for IBM SONAS Storage Node 2. Figure 11-9 shows both hosts in the
cluster.
If the zoning is correct, you can add ports to their storage nodes. Get the worldwide port
name (WWPN) for each SONAS storage node in the name server of the switch. For direct
attached, look at the back of the IBM SONAS storage nodes PCI slot 2 and 4, which have
a label indicating the WWPNs. To add the ports, right-click the host name and select Add
Port as illustrated in Figure 11-10.
6. Map the 4-TB volumes to the cluster so both storage nodes can see the same volumes as
shown in Figure 11-12.
The four volumes are mapped from LUN ID 1 through LUN ID 4 to the IBM SONAS
Gateway cluster as shown in Figure 11-13.
Important: The IBM SONAS Gateway must be ordered as a Gateway and not a normal
SONAS. In this case, XIV is the only storage that the IBM SONAS Gateway can handle.
The installation guide for SONAS Gateway and XIV found at:
http://publib.boulder.ibm.com/infocenter/sonasic/sonas1ic/topic/com.ibm.sonas.doc/
xiv_installation_guide.pdf
Figure 12-1 illustrates attachment of the XIV Storage System with the N Series Gateway.
Figure 12-2 Currently supported N series models and Data ONTAP versions
For the latest information and supported versions, always verify the N series Gateway
interoperability matrix at:
ftp://public.dhe.ibm.com/storage/nas/nseries/nseries_gateway_interoperability.pdf
Figure 12-3 Currently supported Metro and Stretch MetroCluster configurations with XIV
12.4 Zoning
Create zones so that there is only one initiator in each zone. Using a single initiator per zone
ensures that every LUN presented to the N series Gateway has only two paths. It also limits
the registered state change notification (RSCN) traffic in the switch.
The volumes you present from an XIV round up to the nearest increment of 17 GB.
Important: Remember that XIV reports capacity in GB (decimal) and N Series reports in
GiB (Binary). For more information, see 12.5.2, Creating the root volume in XIV on
page 351.
To prepare the XIV Storage System for use with N series Gateway, first create a Storage
Pool using, for instance, the XIV GUI as shown in Figure 12-7.
Tip: No Snapshot space reservation is needed because the XIV snapshots are not
supported with N series Gateways.
N series calculates capacity differently from XIV, and you need to make adjustments to get
the right size. N series GB are expressed as 1000 x 1024 x 1024 bytes, whereas XIV GB are
expressed as either 1000 x 1000 x 1000 bytes or 1024x1024x1024 bytes.
The N series formula is not the same as GB or GiB. Figure 12-8 lists XIV GUI options that
help ensure that you create the correct volume size for other operating systems and hosts.
Note: Inside the XIV GUI, you have several choices in allocation definitions:
If GB units are chosen, a single storage unit is regarded as 109 (1,000,000,000) bytes.
If GiB units are chosen, a single storage unit is regarded as 230 bytes. This is known as
binary notation.
To calculate the size for a minimum N series gateway root volume, use this formula:
<min_size> GB x (1000 x 1024 x 1024) / (1000 x 1000 x1000) = <XIV_size_in_GB>
Because XIV is using capacity increments of about 17 GB, it will automatic set the size to the
nearest increment of 17 GB.
As shown in Figure 12-8, create a volume with the correct size for the root volume in the XIV
pool previously created. Also, create an additional 17-GB dummy volume that can be mapped
as LUN 0. This additional volume might not be needed depending on the particular
environment.
To find the WWPN by using the RLM method, perform these steps:
1. Power on your N series Gateway.
2. Connect to RLM ip using ssh, and log in as naroot.
3. Enter system console.
4. Observe the boot process as illustrated in Example 12-1, and when you see Press CTRL-C
for special boot menu, immediately press Ctrl+C.
Data ONTAP Release 7.3.3: Thu Mar 11 23:02:12 PST 2010 (IBM)
Copyright (c) 1992-2009 NetApp.
Starting boot on Tue Oct 5 17:20:16 GMT 2010
Tue Oct 5 17:20:33 GMT [fci.initialization.failed:error]: Initialization
failed on Fibre Channel adapter 0b.
Tue Oct 5 17:20:39 GMT [diskown.isEnabled:info]: software ownership has been
enabled for this system
Tue Oct 5 17:20:39 GMT [config.noPartnerDisks:CRITICAL]: No disks were
detected for the partner; this node will be unable to takeover correctly
Selection (1-5)?
Make sure that you add both ports as shown in Figure 12-12. If your zoning is right, they
are displayed in the list. If they do not show up, check your zoning.
8. Verify that both ports are connected to XIV by checking the Host Connectivity view in the
XIV GUI as shown in Figure 12-13.
3. Click the 17-GB dummy volume for LUN 0, then map it to LUN 0 by clicking Map as
illustrated in Figure 12-16.
4. Use steps 1-3 to map your N series root volume as LUN 1, also shown in Figure 12-16.
Figure 12-16 XIV Host Mapping view: LUN 0 and LUN 1 mapped correctly
In Windows, these LUNs are reported back to the kernel in response to the SCSI REPORT
LUNS command. Unfortunately, not all vendor storage arrays comply with the standard of
assigning LUN 0 to the controller. Failure to comply with that standard means that the boot
process might not proceed correctly. In certain cases, even with LUN 0 correctly assigned, the
boot LUN cannot be found, and the operating system fails to load. In these cases (without
HBA LUN remapping), the kernel finds LUN 0, but might not be successful in enumerating the
LUNs correctly.
If you are deploying an N series Gateway cluster, you need to map both N series Gateway
root volumes to the XIV cluster group.
Note: If you do not see any disk, make sure that you have Data ONTAP 7.3.3 or later. If you
need to upgrade, follow the N series documentation to perform a netboot update.
Assign the root LUN to the N series Gateway with disk assign <disk name> as shown in
Example 12-3.
Verify the newly assigned disk by entering the disk show command as shown in
Example 12-4.
Data ONTAP Release 7.3.3: Thu Mar 11 23:02:12 PST 2010 (IBM)
Copyright (c) 1992-2009 NetApp.
Starting boot on Wed Oct 6 14:27:17 GMT 2010
Wed Oct 6 14:27:34 GMT [fci.initialization.failed:error]: Initialization
failed on Fibre Channel adapter 0b.
Wed Oct 6 14:27:37 GMT [diskown.isEnabled:info]: software ownership has been
enabled for this system
Remember: Use (4a) flexible root volumes because this option is far more flexible, and
allows more expansion and configuration options than option 4.
4. The N series installs Data ONTAP, and also prompt for environment questions such as IP
address and netmask.
Transfer the right code package to the root volume in directory /etc/software. To transfer the
package from Windows, perform these steps:
1. Start cifs and map c$ of the N series Gateway.
2. Go to the /etc directory and create a folder called software.
Note: Always assign a second LUN to use as the core dump LUN. The size you need
depends on the hardware. Consult the interoperability matrix to find the appropriate size.
For Data ONTAP 2-TB LUN, the XIV size expressed in GB needs to be
2000x(1000x1024x1024) / (1000x1000x1000) = 2097 GB
However, the largest LUN size that can effectively be used in XIV is 2095 GB because XIV
capacity is based on 17-GB increments.
For details about TS7650G ProtecTIER Deduplication Gateway (3958-DD3), see IBM System
Storage TS7650, TS7650G, and TS7610, SG24-7652.
In Figure 13-1, you can see ProtecTIER in a backup solution with XIV Storage System as the
backup storage device. Fibre Channel attachment over switched fabric is the only supported
connection mode.
Application Servers
Windows Unix
IP
Backup Server
TSM
FC
XIV
For details about ProtecTIER, see IBM System Storage TS7650, TS7650G, and TS7610,
SG24-7652, at:
http://www.redbooks.ibm.com/redbooks/pdfs/sg247652.pdf
When attaching the TS7650G ProtecTIER Deduplication Gateway to IBM XIV Storage
System, preliminary conditions must be met. Preparation tasks must be performed in
conjunction with connectivity guidelines already presented in Chapter 1, Host connectivity
on page 1:
Check supported versions and other prerequisites
Physical cabling in place
Define appropriate zoning
Create XIV pool and then volumes
Make XIV host definitions for the ProtecTier Gateway
Map XIV volumes to corresponding ProtecTier Gateway
If a single IBM XIV Storage System is being connected, each Fibre Channel switched fabric
must have six available ports for Fibre Channel cable attachment to IBM XIV Storage System.
Generally, use two connections for each interface module in XIV. Typically, XIV interface
module port 1 is used for Fibre Channel switch 1, and port 3 for switch 2 (Figure 13-2).
When using a partially configured XIV rack, see Figure 1-1 on page 4 to locate the available
FC ports.
XIV
TS7650G ProtecTIER Deduplication Gateway
Figure 13-2 Cable diagram for connecting a TS7650G to IBM XIV Storage System
Each interface module in IBM XIV Storage System has connection with both TS7650G HBAs.
Typical ProtecTIER configuration uses 1:1 zoning (one initiator and one target in each zone)
to create zones. These zones allow the connection of a single ProtecTIER node with a 15
module IBM XIV Storage System with all six Interface Modules. See Example 13-1.
Switch 02:
Zone 01: PT_S6P2, XIV_Module4Port3
Zone 02: PT_S6P2, XIV_Module6Port3
Zone 03: PT_S6P2, XIV_Module8Port3
Zone 04: PT_S7P2, XIV_Module5Port3
Zone 05: PT_S7P2, XIV_Module7Port3
Zone 06: PT_S7P2, XIV_Module9Port3
Tip: In the capacity planning tool for metadata, the fields RAID Type and Drive capacity
show the most optimal choice for an XIV Storage System. The Factoring Ratio number is
directly related to the size of the metadata volumes, and can be estimated using the IBM
ProtecTIER Performance Calculator.
Be sure to take the Max Throughput and Repository Size values into account during the
calculations for both the initial install and future growth.
You must configure the IBM XIV Storage System before the ProtecTIER Deduplication
Gateway is installed on it by an IBM service representative. Perform these steps:
Configure one storage pool for ProtecTIER Deduplication Gateway. Set snapshot space to
zero because creating snapshots on IBM XIV Storage System is not supported with
ProtecTIER Deduplication Gateway.
Configure the IBM XIV Storage System into volumes. Follow the ProtecTIER Capacity
Planning Tool output. The capacity planning tool output gives you the metadata volume
size and the size of the 32 data volumes. Configure a Quorum volume of a minimum of 1
GB as well, in case the solution needs more ProtecTIER nodes in the future.
Map the volumes to ProtecTIER Deduplication Gateway, or, if you have a ProtecTIER
Deduplication Gateway cluster, map the volumes to the cluster.
Tip: Use a Regular Pool and zero the snapshot reserve space. Snapshots and thin
provisioning are not supported when XIV is used with ProtecTIER Deduplication Gateway.
In the example in Figure 13-3 on page 366 with a 79-TB XIV Storage System and a
deduplication Factoring Ratio of 12, the volumes sizes are as follows:
2x 1571-GB volumes for metadata: Make these volumes equal to each other, and nearest
to XIV allocation size, in this case 1583
1x 17 G volume for Quorum (it must be 17 GB because that is the XIV min size)
32 x <Remaining Pool Space available> which is 75440. Dividing 75440 by 32 means that
user data LUNs on the XIV should be 2357 GB each. The XIV 3.0 client GUI makes this
calculation easy for you. Enter the number of Volumes to create, then drag the slider to the
right to fill the entire pool. The GUI automatically calculates the appropriate equivalent
amount.
If you have a ProtecTIER Gateway cluster (two ProtecTIER nodes in a High Availability
solution), perform these steps:
1. Create a cluster group as shown in Figure 13-8.
4. Right-click the cluster and select Add Host as shown in Figure 13-10.
5. Enter the information for the new ProtecTIER host and click Add as shown in
Figure 13-11.
6. Find the WWPNs of the ProtecTIER nodes. WWPNs can be found in the name server of
the Fibre Channel switch. If zoning is in place, they are selectable from the menu.
Alternatively they can also be found in the BIOS of the HBA cards and then entered by
hand in the XUICV GUI.
Figure 13-13 ProtecTIER WWPNs added to XIV Host and Cluster definitions
8. Map the volumes to the ProtecTIER Gateway cluster. In the XIV GUI, right-click the cluster
name or on the host if you have only one ProtecTIER node, and select Modify LUN
Mapping. Figure 13-14 shows you what the mapping view looks like.
Tip: If you have only one ProtecTIER Gateway node, map the volumes directly to the
ProtecTIER gateway node.
Figure 13-15 ProtectTIER Administration window during the setup of the Repository, on XIV
The chapter focuses on the storage-specific aspects of space allocation for database
environments. If I/O bottlenecks show up in a database environment, a performance analysis
of the complete environment might be necessary. Look at database engine, file systems,
operating system, and storage. The chapter also gives hints, tips, and web links for additional
information about the non-storage components of the environment.
Most storage vendors publish recommendations on how to distribute data across the storage
system resources to achieve optimal I/O performance. Unfortunately, the original setup and
tuning cannot usually be kept over the lifetime of an application environment. Because
applications change and storage landscapes grow, traditional storage systems need to be
constantly retuned to maintain optimal performance. One common, less-than-optimal solution
is that additional storage capacity is often provided on a best effort level, and I/O performance
tends to deteriorate.
This aging process that affects application environments on many storage systems does not
occur with the XIV architecture because of these advantages:
Volumes are always distributed across all disk drives.
Volumes can be added or resized without downtime.
Applications get maximized system and I/O power regardless of access patterns.
Performance hotspots do not exist.
Therefore, you do not need to develop a performance-optimized volume layout for database
application environments with XIV. However, it is worth considering some configuration
aspects during setup.
Asynchronous I/O is preferable for an Oracle database on an IBM XIV storage system. The
Oracle database server automatically detects if asynchronous I/O is available on an operating
system. Nevertheless, ensure that asynchronous I/O is configured. Asynchronous I/O is
explicitly enabled by setting the Oracle database initialization parameter DISK_ASYNCH_IO to
TRUE.
For more information about Oracle asynchronous I/O, see Oracle Database High Availability
Best Practices 11g Release 1 and Oracle Database Reference 11g Release 1, available at:
http://www.oracle.com/pls/db111/portal.all_books
The main components of Oracle ASM are disk groups. Each group includes several disks (or
volumes of a disk storage system) that ASM controls as a single unit. ASM refers to the
disks/volumes as ASM disks. ASM stores the database files in the ASM disk groups. These
files include data files, online, and offline redo legs, control files, data file copies, and
Recovery Manager (RMAN) backups. However, Oracle binary and ASCII files, such as trace
files, cannot be stored in ASM disk groups. ASM stripes the content of files stored in a disk
group across all disks in the disk group to balance I/O workloads.
When configuring Oracle database using ASM on XIV, follow these guidelines to achieve
better performance. Theses guidelines also create a configuration that is easy to manage and
use.
Use one or two XIV volumes to create an ASM disk group
Set 8M or 16M Allocation Unit (stripe) size
To achieve optimum database performance and availability, take advantage of the following
unique capabilities of XIV and DB2. This list focuses on the physical aspects of XIV volumes
and how these volumes are mapped to the host.
When creating a database, consider using DB2 automatic storage technology as a simple
and effective way to provision storage for a database. If you use more than one XIV
volume, Automatic storage distributes the data evenly among the volumes. Avoid using
other striping methods such as the logical volume manager of the operating system. DB2
In IBM lab tests, the best performance was achieved with XIV storage system by setting this
variable to 32 or 64 per table space. Example 14-1 shows how to configure DB2 parallel I/O
for all table spaces with the db2set command on AIX.
For more information about DB2 parallelism options, see DB2 for Linux, UNIX, and Windows
Information Center available at:
http://publib.boulder.ibm.com/infocenter/db2luw/v9r7
Both SQL and XIV perform best when host bus adapter (HBA) queue depths are high. SQL
Server applications are generally I/O-intensive, with many concurrent outstanding I/O
requests. As a result, the HBA queue depth needs to be high enough for optimal
performance. By increasing the HBA queue depth, greater amounts of parallel I/O get
distributed as a result of the XIV grid architecture. To maximize the benefits of the XIV parallel
architecture, use a queue depth of 128 for all HBAs attaching to SQL servers.
Remember: Although a higher queue depth in general yields better performance with the
XIV, consider the XIV Fibre Channel (FC) port limitations. The FC ports of the XIV can
handle a maximum queue depth of 1400.
In a SQL environment, there are several ways to achieve parallelism to take advantage of the
XIV grid architecture.
Inter-query parallelism: A single database can accept queries from multiple applications
simultaneously. Each query runs independently and simultaneously.
Intra-query parallelism: Simultaneous processing of parts of a single query using
inter-partition parallelism, intra-partition parallelism, or both.
Intra-partition parallelism: A single query is broken into multiple parts.
Inter-partition parallelism: A single query is broken into multiple parts across multiple
partitions of a partitioned database on a single server or multiple servers. The query
runs in parallel.
Depending on the server hardware and database solution, the maximum degree of
parallelism (MAXDOP) can be configured. For more information about configuring the
MAXDOP option in SQL, see the Microsoft Knowledge Base topic at:
http://support.microsoft.com/kb/329204
SQL backups and restores are I/O intensive. SQL Server uses both I/O parallelism and
intra-partition parallelism when performing backup and restore operations. Backups use
I/O parallelism by reading from multiple database files in parallel, and asynchronously
writing to multiple backup media in parallel.
For batch jobs that are single threaded with time limitations, follow these guidelines:
Use large database buffers to take advantage of prefetching. Allocate as much server
memory as possible.
Run multiple batch jobs concurrently. Even though each batch job takes approximately
the same amount of time, the overall time frame for combined tasks is less.
If possible, break down large batch jobs into smaller jobs. Schedule the jobs to run
simultaneously from the same host or multiple hosts. For example, instead of backing
up data files sequentially, backup multiple data files concurrently.
When performing queries, DDL, DML, data loading, backup, recovery, and replication,
follow guidelines that enhance parallel execution of these tasks.
Microsoft SQL Server automatically tunes many of the server configuration options, so little, if
any, tuning by a system administrator is required. When performance tuning SQL, thoroughly
test the databases. Many factors can influence the outcome, including custom stored
procedures and applications. Generally, avoid using too many performance modifications to
keep the storage configuration simple and streamlined.
For more information, see IBM XIV Storage Tips for Microsoft SQL Server 2008 R2 at:
http://www.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/WP101758
In general, SAP stores all data in one of the following external databases:
DB2 for Linux, UNIX, Windows from IBM
DB2 for IBM i from IBM
MaxDB from SAP
MS SQL Server from Microsoft
Oracle from Oracle
Sybase from SAP
Normally the transaction load of a non-SAP database differs greatly from the load behavior of
an SAP application server. Often non-SAP databases have many random write with 4k blocks
to support small and fast transactions with a low latency. This tendency is in contrast to an
SAP online transaction processing (OLTP) System, which has mostly a 70/30% read/write
workload. SAP Business Warehouse systems are online analytical processing (OLAP)
systems, which have even more sequential shares in the I/O behavior of the database.
Even without a specialized product, you can create a consistent snapshot of a database. To
ensure consistency, the snapshot must include the database, file systems, and storage.
This section gives hints and tips to create consistent storage-based snapshots in database
environments. For more information about f a storage-based snapshot backup of a DB2
database, see the Snapshot chapter of IBM XIV Storage System: Copy Services and
Migration, SG24-7759.
An XIV Consistency Group comprises multiple volumes. Therefore, take the snapshot of all
the volumes at the same time. This action creates a synchronized snapshot of all the
volumes. It is ideal for applications that span multiple volumes that have their transaction logs
on one set of volumes and their database on another.
When creating a backup of the database, synchronize the data so that it is consistent at the
database level as well. If the data is inconsistent, a database restore is not possible because
the log and the data are different and part of the data might be lost.
If consistency groups and snapshots are used to back up the database, database consistency
can be established without shutting down the application. To do so, perform these steps:
1. Suspend the database I/O. With Oracle, an I/O suspend is not required if the backup mode
is enabled. Oracle handles the resulting inconsistencies during database recovery.
2. If the database is in file systems, write all modified file system data back to the storage
system. This process flushes the file systems buffers before creating the snapshots for a
file system sync.
3. Optionally perform file system freeze/thaw operations (if supported by the file system)
before or after the snapshots. If file system freezes are omitted, file system checks are
required before mounting the file systems allocated to the snapshots copies.
4. Use snapshot-specific consistency groups.
A full database restore always requires shutting down the database shut down. If file systems
or a volume manager are used on the operating system level, the file systems must be
unmounted and the volume groups deactivated as well.
The following are the high-level tasks required to perform a full database restore from a
storage-based snapshot:
1. Stop application and shut down the database.
2. Unmount the file systems (if applicable) and deactivate the volume groups.
3. Restore the XIV snapshots.
4. Activate the volume groups and mount the file systems.
5. Recover database, either using complete forward recovery or incomplete recovery to a
certain point in time.
6. Start the database and application.
IBM Tivoli Storage FlashCopy Manager software provides fast application-aware backups and
restores using the advanced snapshot technologies in IBM storage systems.
With optional
TSM backup
Application FlashCopy Manager Offload
System Local System
Application Snapshot
Data Versions
Snapshot
Backup
Sn
a
R e ps ho
sto t
re
DB2 Backup
Oracle to TSM
SAP Restore from TSM
SQL Server
Exchange Server For IBM Storage
SVC
Storwize
V7000
XIV TSM Server
DS8000 Or ThirdParty
DS 3/4/5*
*VSS Integration
IBM Tivoli Storage FlashCopy Manager uses the data replication capabilities of intelligent
storage subsystems to create point-in-time copies. These are application-aware copies
(FlashCopy or snapshot) of the production data. This copy is then retained on disk as a
backup, allowing for a fast restore operation (flashback).
FlashCopy Manager also allows mounting the copy on an auxiliary server (backup server) as
a logical copy. This copy (instead of the original production-server data) is made accessible
for further processing. This processing includes creating a backup to Tivoli Storage Manager
(disk or tape) or doing backup verification functions such as the Database Verify Utility. If a
backup to Tivoli Storage Manager fails, IBM Tivoli Storage FlashCopy Manager can restart
FlashCopy Manager for UNIX and Linux supports the cloning of a Service Advertising
Protocol (SAP) database since release 2.2. In SAP terms, this clone is called a
Homogeneous System Copy. The system copy runs the same database and operating
system as the original environment. Again, FlashCopy Manager uses the FlashCopy or
Snapshot features of the IBM storage system to create a point-in-time copy of the SAP
database. For more information, see 15.3.2, SAP Cloning on page 391.
For more information about IBM Tivoli Storage FlashCopy Manager, see:
http://www.ibm.com/software/tivoli/products/storage-flashcopy-mgr/
For more detailed technical information, visit the IBM Tivoli Storage Manager Version 2.2
Information Center at:
http://publib.boulder.ibm.com/infocenter/tsminfo/v6r2/index.jsp?topic=/com.ibm.its
m.fcm.doc/c_fcm_overview.html
The pre-installation checklist defines hardware and software requirements, and describes the
volume layout for the SAP environment. To ensure a smooth installation of FlashCopy
Manager, all requirements must be fulfilled.
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 385
15.2.1 FlashCopy Manager prerequisites
This section addresses some prerequisites for FlashCopy Manager
DB2 instance volume XIV or local storage DB2 instance directory One dedicated volume
group group
For Oracle, the volume group layout must follow the requirements shown in Table 15-2. The
table space data and redo log directories must be on separate volume groups.
Table space volume XIV Table space files One or more dedicated
groups volume groups
Online redo log XIV Online redo logs, One or more dedicated
volumes groups control files volume groups
The main steps of the FlashCopy Manager installation are shown in Figure 15-2.
additional configuration:
init<SID>.utl
init<SID>.sap
FINISH
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 387
4. Check the summary of the installation wizard as shown in Figure 15-3. Be sure to enter
the correct instance ID of the database.
5. After the installation completes, log in to the server as the database owner and start the
setup_db2.sh script, which asks specific setup questions about the environment.
6. Configure the init<SID>.utl and init<SID>.sap files. This step is only necessary if Tivoli
Storage Manager for Enterprise Resource Planning is installed.
After installing FlashCopy Manager, a profile is required to run FlashCopy Manager properly.
In the following example for a DB2 environment, FlashCopy Manager is configured to back up
to disk only. To create the profile, log in as the db2 database instance owner and run the
setup_db2.sh script on the production system. The script asks several profile-content
questions. The main questions and answers for the XIV storage system are displayed in
Example 15-1 on page 389.
When starting the setup_db2.sh script, enter 1 to configure FlashCopy Manager for backup
only. In Example 15-1 on page 389, part of the XIV configuration is shown and the user input
is indicated in bold.
In this example, the device type is XIV, and the xcli is installed in the /usr/cli directory on
the production system. Specify the IP address of the XIV storage system and enter a valid
Start a disk-only backup with the db2 backup command and the use snapshot clause. DB2
creates a time stamp for the backup image ID that is displayed in the output of the db2 backup
command. The time stamp can also be read with the FlashCopy Manager db2acsutil utility
or the db2 list history DB2 command. This time stamp is required to initiate a restore. For
a disk-only backup, no backup server or Tivoli Storage Manager server is required as shown
in Figure 15-4.
DATA DATA
Snapshot
LOG LOG
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 389
The advantage of the XIV storage system is that snapshot target volumes do not have to be
predefined. FlashCopy Manager creates the snapshots automatically during the backup or
cloning processing.
Example 15-3 shows restore, forward recovery, and activation of the database with the DB2
commands db2 restore, db2 rollforward, and db2 activate.
Tip: Check that enough XIV snapshot space is available for the number of snapshot
versions you want to keep. If the snapshot space is not sufficient, XIV deletes older
snapshot versions.
Snapshot deletions are not immediately reflected in the FlashCopy Manager repository. The
interval of FlashCopy Manager for reconciliation is specified during FlashCopy Manager
setup. It can be checked and updated in the FlashCopy Manager profile. The default of the
RECON_INTERVAL parameter is 12 hours (Example 15-1 on page 389).
You might want to perform system copies for the following reasons:
To create test and quality-assurance systems that are recreated regularly from the
production systems to test new developments with actual production data
To create migration or upgrade systems from the production system before phasing in new
releases or functions into production
To create education systems from a master training set to reset before a new course starts
To create dedicated reporting systems to offload workload from production
SAP defines a system copy as the duplication of an SAP system. Certain SAP parameters
might change in a copy. When you perform a system copy, the SAP SAPinst procedure
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 391
installs all the instances again. However, instead of the database export delivered by SAP, it
uses a copy of the customers source system database to set up the database. Commonly, a
backup of the source system database is used to perform a system copy.
Performing an SAP system copy to back up and restore a production system is a long task
(two or three days). Changes to the target system are applied either manually or supported by
customer-written scripts. Perform a system copy only if you have experience in copying
systems. You also need a good knowledge of the operating system, the database, the ABAP
Dictionary, and the Java Dictionary.
Starting with version 2.2, Tivoli FlashCopy Manager supports the cloning (in SAP terms, a
heterogeneous system copy) of an SAP database. The product uses the FlashCopy or
Snapshot features of IBM storage systems to create a point-in-time copy of the SAP source
database in minutes instead of hours. The cloning process of an SAP database is shown in
Figure 15-6.
The cloning function is useful to create quality assurance (QA) or test systems from
production systems. The renamed clone system can be integrated into the SAP Transport
System that you define for your SAP landscape. Then updated SAP program sources and
other SAP objects can be transported to the clone system for testing purpose.
Production QA Dev/Test
Server Server Server
SAP Transport System
Cloning
Database Database
IBM Solutions Stack delivers
Create DB clone to test SAP
upgrade with production data
IBM can provide a number of preprocessing and postprocessing scripts that automate
important actions. FlashCopy Manager allows you to automatically run these scripts before
and after clone creation, and before the cloned SAP system is started. The pre- and
postprocessing scripts are not part of the FlashCopy Manager software package.
For more information about backup/restore and SAP Cloning with FlashCopy Manager on
UNIX, see the following documents:
Quick Start Guide to FlashCopy Manager for SAP on IBM DB2 or Oracle Database with
IBM XIV Storage System at:
http://www-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/WP101703
Tivoli Storage FlashCopy Manager Version 2.2.x Installation and Users Guide for UNIX
and Linux at:
http://publib.boulder.ibm.com/infocenter/tsminfo/v6r2/index.jsp?topic=/com.ibm.
itsm.fcm.doc/c_fcm_overview.html
This section gives a brief overview of the Microsoft VSS architecture. It also addresses the
requirements, configuration, and implementation of the XIV VSS Provider with Tivoli Storage
FlashCopy Manager for backing up Microsoft Exchange Server data. This combination
provides the tools and information needed to create and manage volume-level snapshots of
Microsoft SQL Server and Microsoft Exchange server data. Tivoli Storage FlashCopy
Manager uses Microsoft VSS in a Windows environment. VSS relies on a VSS hardware
provider. For more information, see 15.5.1, VSS architecture and components on page 395.
Subsequent sections address the installation of the XIV VSS Provider. They also provide
detailed installation and configuration information for the IBM Tivoli Storage FlashCopy
Manager. These sections include usage scenarios.
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 393
Tivoli Storage FlashCopy Manager for Windows is easy to install, configure, and deploy. It
integrates in a seamless manner with any storage system that has a VSS provider. These
systems include IBM System Storage DS3000, DS4000, DS5000, DS8000, IBM SAN Volume
Controller, IBM Storwize V7000, and IBM XIV Storage System.
Figure 15-8 shows the Tivoli Storage FlashCopy Manager management console in the
Windows environment.
Without VSS, if you do not have an online backup solution implemented, you must either stop
or quiesce applications during the backup process. Otherwise, you have an online backup
with inconsistent data and open files that cannot be backed up.
With VSS, you can produce consistent shadow copies by coordinating tasks with business
applications, file system services, backup applications, fast recovery solutions, and storage
hardware. This storage hardware includes the XIV Storage System.
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 395
For SQL data, Microsoft SQL Server contains the writer components (SqlServerWriter).
These components are installed with the SQL Server software, and require the following
minor configuration tasks:
Set the SqlServerWriter service to automatic. This setting allows the service to start
automatically when the system is rebooted.
Start the SqlServerWriter service.
Provider
This application produces the shadow copy and manages its availability. It can be a
system provider such as the one included with the Microsoft Windows operating system. It
can also be a software provider, or a hardware provider such as the one available with the
XIV storage system.
For XIV, you must install and configure the IBM XIV VSS Provider.
VSS uses the following terminology to characterize the nature of volumes participating in a
shadow copy operation:
Persistent
This shadow copy remains after the backup application completes its operations. This type
of shadow copy also remains intact after system reboots.
Non-persistent
This temporary shadow copy remains until the backup application needs it to copy the
data to its backup repository.
Transportable
This shadow copy volume is accessible from a secondary host so that the backup can be
moved to another host. Transportable is a feature of hardware snapshot providers. On an
XIV, you can mount a snapshot volume to another host.
Source volume
This volume contains the data to be shadow copied. These volumes contain the
application data.
Target or snapshot volume
This volume retains the shadow-copied storage files. It is an exact copy of the source
volume at the time of backup.
When the snapshot is logically completed, VSS allows writes to resume and notifies the
requestor that the backup is completed successfully. The backup volumes are mounted, but
are hidden and read-only. The backup therefore is ready to be used when a rapid restore is
requested. Alternatively, the volumes can be mounted to a separate host and used for
application testing or backup to tape.
At the time of writing, the XIV VSS Provider 2.3.1 version was available. A Windows 2008
64-bit host system was used for the tests. For more information about the system
requirements, see the IBM VSS Provider - Xprov Release Notes. There is a chapter about the
system requirements.
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 397
The XIV VSS Hardware Provider 2.3.1 version and release notes can be downloaded at:
http://www.ibm.com/support/entry/portal/Downloads
The installation of the XIV VSS Provider is a straightforward Windows application installation:
1. Locate the XIV VSS Provider installation file, also known as the xProv installation file. If
the XIV VSS Provider 2.3.1 is downloaded from the Internet, the file name is
xProvSetup-2.3.1-x64.exe. Run the file to start the installation.
Tip: Uninstall any previous versions of the XIV VSS xProv driver. The 2.3.1 release of
XIV VSS provider does not support upgrades.
3. The License Agreement window is displayed. To continue the installation, accept the
license agreement.
4. Specify the XIV VSS Provider configuration file directory and the installation directory.
Keep the default directory folder and installation folder, or change them to meet your
needs.
6. A Confirm Installation window is displayed. If required you can go back to make changes
or confirm the installation by clicking Next.
7. After the installation is complete click Close to exit.
2. Right-click the Machine Pool Editor. The New System window displays. Provide specific
information regarding the XIV Storage System IP addresses and user ID and password
with admin privileges.
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 399
3. In the window shown in Figure 15-13, click New System.
4. The Add Storage System window shown in Figure 15-14 is displayed. Enter the user name
and password of an XIV user with administrator privileges (storageadmin role) and the
primary IP address of the XIV Storage System. Click Add.
5. You are now returned to the VSS MachinePool Editor window. The VSS Provider collected
additional information about the XIV storage system, as illustrated in Figure 15-15.
The XIV VSS Provider configuration is now complete, and you can close the Machine Pool
Editor window. Repeat steps 3-5 for any additional XIV Storage Systems. You must add a
second XIV system if you make snapshots from mirrored LUNs with Enable Replicated
Snapshots selected. A second system is need because the VSS provider must know both
the master and subordinate mirror sites.
The Windows server is now ready to perform snapshot operations on the XIV Storage
System. For more information about completing the VSS setup, see your application
documentation.
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 401
xProv tracing
If you detect errors which might be related to a problem in the VSS Provider, you can change
the debugging level from normal to debug. Navigate to the installation directory of xProv
software, which is normally in the folder C:\Program Files\IBM\IBM XIV Provider for
Microsoft Windows Volume Shadow Copy Service\etc. Perform the following steps to edit the
file log4net_conf.xml:
1. Change line 3 to read level="INFO" to level="DEBUG".
2. Save and restart the xProv service and run the backup again.
3. Send back the log file in c:\windows\temp\xProvDotNet\xProvDotNet.log to IBM support.
Diskshadow is a requester, and communicates with the VSS writer and a VSS provider such
as XIV to coordinate the snapshot process.
During the actual volume shadow process, diskshadow requests XIV snapshots of the
defined volumes. The generic script shown in Example 15-6 details the syntax necessary to
create diskshadow snapshots. Start the script by entering this command:
diskshadow /s vss_backup.txt
Tip: By using the transportable option, the snapshot can be imported on a different SQL
host and used for testing, backups, or for data mining.
In the XIV GUI shown in Figure 15-16, the diskshadow-initiated snapshots are different from
XIV-initiated snapshots. The diskshadow snapshots have a different naming convention. The
prefix begins with VSS- for obvious reasons and does not contain the suffix .snapshot_0000x.
However, the diskshadow-initiated snapshots are locked as read-only just like XIV-initiated
snapshots. They are locked due to the transportable option used in the script example.
Important: Always thoroughly test scripts in a pilot environment before implementing them
in a production environment.
The procedure gives database administrators another option in their backup arsenal. It is best
to use VSS-based snapshots in conjunction with other traditional or preferred backup
methods. Using these snapshots provides the ultimate flexibility when it comes to rolling
forward uncommitted transactions or rolling backwards through committed transactions
during recoveries.
For more information about diskshadow, see the Microsoft TechNet topic at:
http://technet.microsoft.com/en-us/library/cc772172(WS.10).aspx
2. Enter the user name and password of an XIV user with administrator privileges
(storageadmin role) and the primary IP address of the XIV Storage System.
3. Select Enable Replicated Snapshots and click Add.
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 403
4. The VSS MachinePool Editor window opens. Provide the complete information for the
second XIV, as illustrated in Figure 15-18.
Figure 15-19 shows that this snapshot was taken at the primary (master) site.
Figure 15-20 shows that the VSS Provider created a snapshot with the same name on the
subordinate site.
If you lose the master site as a result of a disaster or outage at the datacenter, all necessary
snapshots are at the surviving site.
The Tivoli Storage FlashCopy Manager installation and configuration wizards guides you
through the installation and configuration steps. After you run the setup and configuration
wizards, your computer is ready to take snapshots.
Tivoli Storage FlashCopy Manager provides the following wizards for installation and
configuration tasks:
Setup wizard
Use this wizard to install Tivoli Storage FlashCopy Manager on your computer.
After it is installed, Tivoli Storage FlashCopy Manager must be configured for VSS snapshot
backups. Use the local configuration wizard for that purpose. These tasks include selecting
the applications to protect, verifying requirements, provisioning, and configuring the
components required to support the selected applications.
2. A dialog window is displayed as shown in Figure 15-22. Select the Exchange Server to
configure and click Next.
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 405
Tip: Click Show System Information to display the basic information about your host
system.
Select the check box at the bottom if you do not want the local configuration wizard to start
automatically the next time that the Tivoli Storage FlashCopy Manager Management
Console windows starts.
3. The Requirements Check window is displayed as shown in Figure 15-23. The systems
checks that all prerequisites are met.
If any requirement is not met, the configuration wizard does not proceed to the next step.
You might need to upgrade components to fulfill the requirements. The requirements
check can be run again by clicking Re-run after they are fulfilled. When the check
completes successfully, click Next.
Remember: By default, details are hidden. Change this setting by clicking Show
Details or Hide Details.
5. The completion window shown in Figure 15-25 is displayed. To run a VSS diagnostic
check, ensure that the corresponding check box is selected and click Finish.
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 407
6. The VSS Diagnostic window is displayed. Verify that any volume that you select can
perform an XIV snapshot using VSS by selecting it and clicking Next (Figure 15-26).
Tip: Any previously taken snapshots can be seen by clicking Snapshots, which
refreshes the list and shows all of the existing snapshots.
8. A completion window is displayed with the results. When done, click Finish.
Tip: Microsoft SQL Server can be configured the same way as Microsoft Exchange
Server to perform XIV VSS snapshots for Microsoft SQL Server using Tivoli Storage
FlashCopy Manager.
In the following example of performing a VSS snapshot backup of Exchange data, the
following setup was used:
Windows 2008 64-bit
Exchange 2007 Server
XIV Host Attachment Kit 1.0.4
XIV VSS Provider 2.0.9
Tivoli Storage FlashCopy Manager 2.0
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 409
Microsoft Exchange Server XIV VSS Snapshot backup
On the XIV Storage System, a single volume is mapped to the host system as illustrated in
Figure 15-28. On the Windows host system, the volume is initialized as a basic disk and
assigned the drive letter G. The G drive is formatted as NTFS, and a single Exchange Server
storage group with mailboxes created on that drive.
Tivoli Storage FlashCopy Manager is already configured and tested for XIV VSS snapshot as
shown in 15.7, Installing Tivoli Storage FlashCopy Manager for Microsoft Exchange on
page 404. To review the Tivoli Storage FlashCopy Manager configuration settings, use the
command shown in Example 15-8.
Example 15-8 Reviewing Tivoli Storage FlashCopy Manager for Mail settings
C:\Program Files\Tivoli\TSM\TDPExchange>tdpexcc query tdp
BACKUPDESTination................... LOCAL
BACKUPMETHod........................ VSS
BUFFers ............................ 3
BUFFERSIze ......................... 1024
DATEformat ......................... 1
LANGuage ........................... ENU
LOCALDSMAgentnode................... sunday
LOGFile ............................ tdpexc.log
LOGPrune ........................... 60
MOUNTWait .......................... Yes
NUMberformat ....................... 1
REMOTEDSMAgentnode..................
As explained earlier, Tivoli Storage FlashCopy Manager does not use (or need) a Tivoli
Storage Manager server to perform a snapshot backup. Run the query tsm command as
shown in Example 15-9. The output does not show a Tivoli Storage Manager service.
FLASHCOPYMANAGER is shown instead in the NetWork Host Name of Server field. Tivoli Storage
FlashCopy Manager creates a virtual server instead of using a Tivoli Storage Manager Server
to perform a VSS snapshot backup.
Example 15-10 shows what options are configured and used for Tivoli Storage Manager
Client Agent to perform VSS snapshot backups.
*======================================================================*
* TCP/IP Communication Options *
*======================================================================*
COMMMethod TCPip
TCPSERVERADDRESS FlashCopymanager
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 411
TCPPort 1500
TCPWindowsize 63
TCPBuffSize 32
Before performing any backup, ensure that VSS is properly configured for Microsoft
Exchange Server and that the DSMagent service is running (Example 15-11).
STG3G_XIVG2_BAS
Circular Logging - Disabled
Replica - None
Recovery - False
2nd MailBox Online
Mail Box1 Online
The test Microsoft Exchange Storage Group is on drive G:\, and it is called STG3G_XIVG2_BAS.
It contains two mailboxes:
Mail Box1
2nd MailBox
Note that a disk drive is not specified here. Tivoli Storage FlashCopy Manager finds out which
disk drives to copy with the snapshot while backing up a Microsoft Exchange Storage Group.
This process is the advantage of an application-aware snapshot backup process.
To see a list of the available VSS snapshot backups, issue a query command as shown in
Example 15-13.
Querying FlashCopy Manager server for a list of database backups, please wait...
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 413
Backup List
-----------
To show that the restore operation is working, the 2nd Mailbox mail box was deleted as
shown in Example 15-14.
To perform a restore, all the mailboxes must be unmounted first. A restore is done at the
volume level, which is called an instant restore. When that is complete, the recovery operation
runs, applying all the logs and mounting the mail boxes. The process is shown in
Example 15-15.
Starting snapshot restore process. This process may take several minutes.
C:\Program Files\Tivoli\TSM\TDPExchange>
Consideration: Instant restore is at the volume level. It does not show the total number of
files examined and completed like a normal backup process does.
To verify that the restore operation worked, open the Exchange Management Console and
check that the storage group and all the mailboxes are mounted. Furthermore, verify that the
2nd Mailbox.edb file exists.
For more information, see the Tivoli Storage FlashCopy Manager: Installation and Users
Guide for Windows, SC27-2504, or Tivoli Storage FlashCopy Manager for AIX: Installation
and Users Guide, SC27-2503.
The latest information about the Tivoli Storage FlashCopy Manager is available at:
http://www.ibm.com/software/tivoli
Chapter 15. Snapshot Backup/Restore Solutions with XIV and Tivoli Storage FlashCopy Manager 415
416 IBM XIV Storage System: Host Attachment and Interoperability
A
The goal of this appendix is only to give you enough information to quickly install, configure,
and experiment with Site Recovery Manager (SRM). It is not meant as a guide on how to
deploy SRM in a real production environment,
VMware SRM uses the replication (mirroring function) capabilities of the underlying storage to
create a copy of the data to a second location (a backup data center). This process ensures
that two copies of the data are always available. If the one currently used by production fails,
production can be switched to the other copy.
In a normal production environment, the virtual machines (VMs) run on ESX hosts and use
storage systems in the primary datacenter. Additional ESX servers and storage systems
stand by in the backup datacenter. The mirroring functions of the storage systems maintain a
copy of the data on the storage device at the backup location. In a failover scenario, all VMs
are shut down at the primary site. The VMs are then restarted on the ESX hosts at the backup
datacenter, accessing the data on the backup storage system.
VMware SRM can automate these tasks and perform the necessary steps to fail over the
complete virtual environments with just one click. This automation saves time, eliminates user
errors, and helps provide detailed documentation of the disaster recovery plan.
SRM can also perform a test of the failover plan by creating an additional copy of the data on
the backup system. It can start the virtual machines from this copy without connecting them to
any network. This configuration allows you to test recovery plans without interrupting
production systems.
At a minimum, an SRM configuration consists of two ESX servers, two vCenters, and two
storage systems, one each at the primary and secondary locations. The storage systems are
configured as a mirrored pair relationship. Ethernet connectivity between the two locations is
required for the SRM to function properly.
For more information about the concepts, installation, configuration, and usage of VMware
Site Recovery Manager, see the VMware product site at:
http://www.vmware.com/support/pubs/srm_pubs.html
This appendix provides specific information about installing, configuring, and administering
VMware Site Recovery Manager in conjunction with IBM XIV Storage Systems.
At the time of this writing, versions 1.0, 1.0 U1, and 4.0 of Storage Replication Agent for
VMware SRM server are supported with XIV Storage Systems.
For more information, see the VMware SRM documentation as previously noted, and in
particular to the VMware vCenter Site Recovery Manager Administration Guide at:
http://www.vmware.com/pdf/srm_admin_4_1.pdf
See Chapter 1, Host connectivity on page 1 and Chapter 9, VMware ESX/ESXi host
connectivity on page 263 for information about implementing the first six bullets.
Tip: Use single initiator zoning to zone ESX host to all available XIV interface modules.
Steps to meet these prerequisites are described in the next sections of this chapter. The
examples show how to set up a simple SRM server installation in your environment.
After you meet all of the prerequisites, you can test your recovery plan. After a successful
test, you can perform a fail-over scenario for your primary site. Be prepared to run the virtual
machines at the recovery site for an indefinite amount of time because VMware SRM server
does not support automatic fail-back operations.
Both of these options require downtime for the virtual machines involved.
The SRM server needs to have its own database for storing recovery plans, inventory
information, and similar data. SRM supports the following databases:
IBM DB2
Microsoft SQL
Oracle
The SRM server has a set of requirements for the database implementation. Some of these
requirements are general and do not depend on the type of database used, but others are
The SRM server database can be on the same server as vCenter, on the SRM server host, or
on a separate host. The location depends on the architecture of your IT landscape and on the
database that is used.
Information about compatibility for SRM server versions can be found at the following
locations:
Version 4.0 and later:
http://www.vmware.com/pdf/srm_compat_matrix_4_x.pdf
Version 1.0 update 1:
http://www.vmware.com/pdf/srm_101_compat_matrix.pdf
Version 1.0:
http://www.vmware.com/pdf/srm_10_compat_matrix.pdf
The Microsoft SQL Express database is available at no additional cost for testing and
development purposes. It is available for download from the Microsoft website at:
http://www.microsoft.com/downloads/en/details.aspx?FamilyID=3181842A-4090-4431-ACD
D-9A1C832E65A6&displaylang=en
The graphical user interface for the database can be downloaded for at no additional cost
from the Microsoft website at:
http://www.microsoft.com/downloads/details.aspx?FamilyId=C243A5AE-4BD1-4E3D-94B8-5
A0F62BF7796&DisplayLang=en
Further Information: For specific requirements and details about installing and
configuring the database application, see the database vendor and VMware
documentation for SRM.
2. The installation wizard starts. Proceed through the prompts until you reach the Feature
Selection window shown in Figure A-2. Connectivity Components must be selected for
installation. Click Next.
4. The Authentication Mode window is displayed as shown in Figure A-4. Select Windows
Authentication Mode for a simple environment. Depending on your environment and
needs, you might need to choose another option. Click Next.
6. The Error and Usage Report Settings window is displayed as shown in Figure A-6.
Select whether you want to report errors to Microsoft Corporation and click Next.
8. After the installation process is complete, the dialog window shown in Figure A-8 is
displayed. Click Next to complete the installation procedure.
Figure A-10 Starting installation for Microsoft SQL Server Management Studio Express
After clicking the file, the installation wizard starts. Proceed with the required steps to
complete the installation.
The Microsoft SQL Server Management Studio Express software must be done installed
at all locations involved in your continuity and disaster recovery solution.
7. Configure one vCenter database and one SRM database for each site. The examples
provide instructions for the vCenter database. Repeat the process for the SRM server
database and the vCenter database at each site. Start Microsoft SQL Server Management
Studio Express by clicking Start All programs Microsoft SQL Server 2005, and
then click SQL Server Management Studio Express (Figure A-13).
9. After successful login, the MS SQL Server Management Suite Express main window is
displayed (Figure A-15). In this window, use configuration tasks to create databases and
logins. To create databases, right-click Databases and select New database.
10.Enter the information for the database name, owner, and database files. In this example,
only the database name is set, leaving all others parameters at their default values. Click
OK to create your database.
12.Right-click the subfolder logins and select new login in the window (Figure A-17). Enter the
user name, type of authentication, default database, and default code page. Click OK.
Repeat this action for the vCenter and SRM servers databases.
Figure A-18 Granting the rights on a database for the login created
You are ready to start configuring ODBC data sources for the vCenter and SRMDB
databases on the server you plan to install them on.
15.To configure ODBC data stores, click Start in the Windows desktop task bar and select
Administrative Tools Data Source (ODBC).
17.The Create New Data Source window opens as shown in Figure A-20. Select SQL Native
Client and click Finish.
18.The window shown in Figure A-21 on page 432 opens. Enter information for your data
source like the name, description, and server for the vcenter database. In the example, the
parameters are set as follows:
Name parameter to vcenter
Description parameter to database for vmware vcenter
Server parameter to SQLEXPRESS.
Click Next.
19.The window shown in Figure A-22 opens. Select With Integrated Windows
Authentication and Connect to SQL Server to obtain default settings to the
additional configuration options, then click Next.
21.The window shown in Figure A-24 is displayed. Select Perform translation for the
character data and then click Finish.
23.The window shown in Figure A-26 indicates that the test completed successfully. Click OK
to return to the previous window, and then click Finish.
Install and configure databases on all sites that you plan to include into your business
continuity and disaster recovery solution.
Now you are ready to proceed with the installation of vCenter server, vCenter client, SRM
server, and SRA agent.
Further Information: For detailed information about vCenter server installation and
configuration, see the VMware documentation. This section includes only common, basic
information for a simple installation used to demonstrate SRM server capabilities with the
IBM XIV Storage System.
4. In the window shown in Figure A-29, type the password for the system account and click
Next.
Figure A-30 Selecting Linked Mode options for the vCenter server
6. In the window shown in Figure A-31, you can change default settings for ports used for
communications by the vCenter server. For most implementations, keep the default
settings. Click Next.
8. The window shown in Figure A-33 indicates that the system is now ready to install
vCenter. Click Install.
You need to install vCenter server on all sites that you plan to include as part of your business
continuity and disaster recovery solution.
Note: For detailed information about vSphere Client, and complete installation and
configuration instructions, see the VMware documentation. This chapter includes only
basic information about installing the vSphere Client and using it to manage the SRM
server.
Locate the vCenter server installation file (either on the installation CD or a copy you
downloaded from the Internet). Running the installation file displays the vSphere Client
installation wizard welcome dialog. Follow the installation wizard instructions to complete the
installation. You need to install vSphere Client on all sites that you plan to include in your
business continuity and disaster recovery solution.
3. Add the new datacenter under control of the newly installed vCenter server. In the main
vSphere Client window, right-click the server name and select New Datacenter as shown
in Figure A-36.
5. The Add Host wizard is started. Enter the name or IP address of the ESX host, user name
for the administrative account on this ESX server, and the account password
(Figure A-38). Click Next.
7. Verify the settings discovered for the specified ESX host as shown in Figure A-40. Check
the information, and, if all is correct, click Next.
9. Select a location for the newly added ESX server as shown in Figure A-42 and click Next.
Figure A-42 Selecting the location in the vCenter inventory for the virtual machines of the host
11.You return to the vSphere Client main window as shown in Figure A-44.
Figure A-44 Presenting inventory information about ESX server in the vCenter database
Repeat all these steps for all the vCenter servers you want to include into your business
continuity and disaster recovery solution.
4. Provide the vCenter server IP address, vCenter server port, vCenter administrator user
name, and password for the administrator account (Figure A-46). Click Next.
6. Select Automatically generate certificate as shown in Figure A-48, and click Next.
Tip: If your vCenter servers are using NON-default (that is, self signed) certificates,
select Use a PKCS#12 certificate file. For more information, see the VMware vCenter
Site Recovery Management Administration Guide at:
http://www.vmware.com/pdf/srm_admin_4_1.pdf
8. The window shown in Figure A-50 asks for general parameters pertaining to your SRM
installation. Provide the location name, administrator email, additional email, local host IP
address or name, and the ports to be used for connectivity. When done, click Next.
Figure A-50 General SRM server settings for the installation location
10.The next window informs you that the installation wizard is ready to proceed as shown in
Figure A-52. Click Install to start the installation process.
You need to install SRM server on each protected and recovery site that you plan to include
into your business continuity and disaster recovery solution.
4. The Plug-in Manager window opens. Under the category Available plug-ins. right-click
vCenter Site Recovery Manager Plug-in and select Download and Install as shown in
Figure A-54.
5. The vCenter Site Recovery Manager Plug-in wizard is started. Follow the wizard
guidelines to complete the installation.
You need to install SRM plug-in on each protected and recovery site that you plan to include
in your business continuity and disaster recovery solution.
For more information about connecting ESX hosts to the IBM XIV Storage, see Chapter 9,
VMware ESX/ESXi host connectivity on page 263.
Create a storage pool on the IBM XIV Storage System at the recovery site. The new storage
pool contains the replicas of the ESX host data stores that are associated with virtual
machines that you plan to protect.
Remember: Configure a snapshot size of at least 20 percent of the total size of the
recovery volumes in the pool. For testing failover operations that can last several days,
increase the snapshot size to half the size of the recovery volumes in the pool. For
longer-term or I/O intensive tests, the snapshot size might have to be the same size as the
recovery volumes in the pool.
For the information about IBM XIV Storage System LUN mirroring, see the IBM Redbooks
publication IBM XIV Storage System: Copy Services and Migration, SG24-7759.
At least one virtual machine for the protected site must be stored on the replicated volume
before you can start configuring SRM server and SRA adapter. In addition, avoid replicating
swap and paging files.
3. Click Site Recovery at the bottom of the main vSphere Client window as shown in
Figure A-57.
Figure A-57 Running the Site Recovery Manager from the vCenter Client menu
5. The Connection to Remote Site window displays. Type the IP address and ports for the
remote site as shown in Figure A-59. Click Next.
7. In the window shown in Figure A-61, enter the user name and password to be used for
connecting at the remote site. Click Next.
8. A remote vCenter server certificate error is displayed as shown in Figure A-62. Ignore this
error message and click OK.
10.In the main SRM server configuration window (Figure A-58 on page 452), click Configure
for Array Managers.
11.In the window shown in Figure A-64, click Add.
13.Click OK to be returned to the previous window where you can observe the remote XIV
paired with local XIV storage system. Click Next.
14.Provide connectivity information for managing your auxiliary storage system as shown in
Figure A-66. Click Next.
18.From the main SRM server configuration window, click Create next to the Protection
Group table.
20.Select data stores to be associated with your protection group as shown in Figure A-70.
Click Next.
6. Select protection groups from your protected site for inclusion in the recovery plan as
shown in Figure A-74, then click Next.
7. The Response Times window is displayed as shown in Figure A-75. Type the values for
your environment or leave the default values, then click Next.
Figure A-76 Configuring the networks which would be used for failover
9. Select the virtual machines that are suspended at the recovery site when fail-over occurs
as shown in Figure A-77. Make your selection and click Finish.
Figure A-77 Selecting virtual machines which would be suspended on recovery site during failover
All steps required to install and configure a simple, proof-of-concept SRM server configuration
are complete.
The publications listed in this section are considered particularly suitable for a more detailed
discussion of the topics covered in this book.
IBM Redbooks
For information about ordering these publications, see How to get Redbooks on page 462.
Note that some of the documents referenced here may be available in softcopy only.
IBM System Storage TS7650, TS7650G, and TS7610, SG24-7652
IBM System z Connectivity Handbook, SG24-5444
IBM XIV Storage System: Architecture, Implementation, and Usage, SG24-7659
IBM XIV Storage System: Copy Services and Migration, SG24-7759
Implementing the IBM System Storage SAN Volume Controller V5.1, SG24-6423
Introduction to Storage Area Networks, SG24-5470
IBM PowerVM Virtualization Introduction and Configuration, SG24-7940
Other publications
These publications are also relevant as further information sources:
IBM XIV Storage System Application Programming Interface, GC27-3916
IBM XIV Storage System Host Attachment Guide: Host Attachment Kit for AIX,
GA32-0643
IBM XIV Storage System Host Attachment Guide: Host Attachment Kit for HPUX,
GA32-0645
IBM XIV Storage System Host Attachment Guide: Host Attachment Kit for Linux,
GA32-0647
IBM XIV Storage System Host Attachment Guide: Host Attachment Kit for Solaris,
GA32-0649
IBM XIV Storage System Host Attachment Guide: Host Attachment Kit for Windows,
GA32-0652
IBM XIV Storage System Planning Guide, GC27-3913
IBM XIV Storage System Pre-Installation Network Planning Guide for Customer
Configuration, GC52-1328-01
IBM XIV Storage System: Product Overview, GC27-3912
IBM XIV Remote Support Proxy Installation and Users Guide, GA32-0795
IBM XIV Storage System User Manual, GC27-3914
IBM XIV Storage System XCLI Utility User Manual, GC27-3915
G I
General Parallel File System (GPFS) 332 I/O operation 220, 222
given storage system path failover 280 I/O request 221222
GPFS See General Parallel File System (GPFS) maximum number 221
GPT See GUID Partition Table (GPT) IBM i 105
Grand Unified Bootloader (GRUB) 137 best practices 221
GRUB See Grand Unified Bootloader (GRUB) queue depth 220
GUID Partition Table (GPT) 69 IBM i operating (I/O) 215217
IBM Power System 216
IBM SONAS
H Gateway 332334
HA See High Availability (HA) cluster 338, 341
hard zone 18 code 342
hardware accelerated initialization 298 component 342
Hardware Accelerated Move 298 Gateway cluster 338
Hardware Assisted Locking 298 Gateway Storage Node 333, 335
Hardware Management Console (HMC) 216 Storage Node 340
POWER6 222 Storage Node 2 340
HBA See host bus adapter (HBA) version 1.1.1.0-x 333
High Availability (HA) 265, 267 IBM System Storage
Host Attachment Kit 101 Interoperability Center 13, 27
attaching XIV volumes 106 IBM Tivoli Storage FlashCopy Manager 393
installing 192, 205 IBM XIV Storage System 12, 93, 177, 191, 264
Linux utilities 131 administrator 2
package 205 and ProtecTIER Deduplication Gateway 361
portable 150 architecture 222
utilities 197 attaching 204
Windows installation 49 to IBM SONAS Gateway 331
XIV 1.0.4 409 to N series Gateway 343
Host Attachment Wizard 49 XenServer hosts 326
host bus adapter (HBA) 2, 14, 268 DMP multipathing 204
driver 19, 103 end-to-end support 264
drivers 269, 276 engineering team 264
Fibre Channel (FC) 99, 101 FC HBA 37
iSCSI 12 GUI 20, 3133, 41
kernel levels 288 I/O performance 222
queue depth 377 iSCSI IPs 37
supported 13 iSCSI IQN 37
Index 465
queue depth 377 P
Microsoft Systems Operation Manager (SCOM) 10 patch panel 3
Microsoft Volume Shadow Copy Services (VSS) 383, path control module (PCM) 153
394 Path Selection Plug-In (PSP) 280
architecture 395 PCM See path control module (PCM)
provider 393, 396, 401 physical disk 221
requestor 395 queue depth 221
service 395 Pluggable Storage Architecture (PSA) 280281
writer 395 portable Host Attachment Kit 150
Microsoft Windows 2008 R2 cluster 66 POST See power-on self test (POST)
mktcpip 226 power-on self test (POST) 137
modinfo 102 PowerVM 216
modprobe 102 Enterprise Edition 217
Most Recently Used (MRU) 272 Express Edition 216
mpathconf 95, 119 Live Partition Mobility 217
MPIO See multi-path I/O (MPIO) Standard Edition 217
MRU See Most Recently Used (MRU) prefetching 378
MSDSM 47 ProtecTIER 362
MTU See maximum transmission unit (MTU) provider 393, 396
multipath 221 PSA See Pluggable Storage Architecture (PSA)
multipath device 95, 110, 115, 124, 127 PSP See Path Selection Plug-In (PSP)
boot Linux 142 pvlinks 180
multi-path I/O (MPIO) 46, 153 Python 10
commands 156 engine 49
multipathing 12, 145, 147
Q
N QLogic
N series Gateway 344 BIOS 138
N10116 395 queue depth 47, 153, 211, 220222, 288, 293, 377
NAA See Network Address Authority (NAA) following types 220
NAS See Network Attached Storage (NAS) queue_depth 153
Native Multipathing (NMP) 280 quiesce 397
Network Address Authority (NAA) 21
Network Attached Storage (NAS) 332, 344
Network File System (NFS) 344 R
NFS See Network File System (NFS) Red Hat Enterprise Linux (RH-EL) 94
NMP See Native Multipathing (NMP) 5.2 192
Node Port ID Virtualization (NPIV) 216, 218 Redbooks Web site 462
NPIV See Node Port ID Virtualization (NPIV) Contact us xiv
NTFS 69 redirect-on-write 396
reference materials 94
registered state change notification (RSCN) 18, 348
O Remote Login Module (RLM) 352
OLAP See online analytical processing (OLAP) remote mirroring 13
OLTP See online transaction processing (OLTP) remote port 110, 112113, 125
online analytical processing (OLAP) 379 sysfs structure 125
online transaction processing (OLTP) 379 unit_remove meta file 126
only specific (OS) 203, 205 request for price quotation (RPQ) 12
operating system (OS) requestor 395, 397
boot loader 137 RH-EL See Red Hat Enterprise Linux (RH-EL)
diagnostic information 132 RLM See Remote Login Module (RLM)
volume manager 374 rootvg 226
original data 396 Round Robin 63
exact read-only copy 396 round_robin 157
full copy 396 round-robin 0 118, 122
OS level 204 RSCN See registered state change notification (RSCN)
unified method volume management 204
OS Type 37
S
same LUN 13, 212, 226
Index 467
device 233 Z
HBA 99 zoning 18, 20, 38
virtual tape 217, 362
virtualization 216
virtualization management (VM) 215216, 218219
virtualization task 97, 114
VM See virtualization management (VM)
VMware ESX
3.5 20
3.5 host 268
4 280, 290, 295
server 264265
server 3.5 268
VMware Site Recovery Manager
IBM XIV Storage System 266
Volume Group 158
VSS See Microsoft Volume Shadow Copy Services (VSS)
vssadmin 401
vStorage API Array Integration (VAAI) 266, 292, 298
vxdisk 207
vxdiskadm 182
vxdiskconfig 207
vxdmpadm 211
VxVM 182
W
WAFL See Write Anywhere File Layout (WAFL)
worldwide ID (WWID)WWID See worldwide ID (WWID)
worldwide node name (WWNN) 111
worldwide port name (WWPN) 78, 20, 99, 340, 352
Write Anywhere File Layout (WAFL) 350
Write Barrier 98
writer 395, 397
WWNN See worldwide node name (WWNN)
WWPN See worldwide port name (WWPN)
X
XCLI See Extended Command Line Interface (XCLI)
XenCenter 322
XenConverter 322
XenMotion 322
XenServer
hypervisor 322
XIV See IBM XIV Storage System
XIV VSS Provider 393
configuration 399
XIV VSS provider 397
xiv_attach 10, 107
xiv_detach 10
xiv_devlist 11, 109110, 131, 232
xiv_diag 11, 132, 197
xiv_fc_admin 11
xiv_iscsi_admin 12
XIVDSM 47
XIVmscsAgent 75, 77
XIVTop 20
xpyv 49
Integrate with DB2, This IBM Redbooks publication provides information for
attaching the IBM XIV Storage System to various host operating
INTERNATIONAL
VMware ESX,
system platforms, including IBM i. The book also addresses TECHNICAL
Microsoft HyperV, and
using the XIV storage with databases and other storage-oriented SUPPORT
SAP
application software, including: ORGANIZATION
Get operating system IBM DB2
specifics for host side VMware ESX
tuning Microsoft HyperV
SAP BUILDING TECHNICAL
Use with IBM i, The book also addresses combining the XIV Storage System with INFORMATION BASED ON
PRACTICAL EXPERIENCE
SONAS, N series, and other storage platforms, host servers, or gateways, including
ProtecTIER IBM SONAS, IBM N Series, and IBM ProtecTIER. It is intended for IBM Redbooks are developed
administrators and architects of enterprise storage systems. by the IBM International
The goal is to give an overview of the versatility and Technical Support
Organization. Experts from
compatibility of the XIV Storage System with various platforms IBM, Customers and Partners
and environments. from around the world create
The information presented here is not meant as a replacement timely technical information
based on realistic scenarios.
or substitute for the Host Attachment kit publications. It is meant Specific recommendations
as a complement and to provide readers with usage guidance are provided to help you
and practical illustrations. implement IT solutions more
effectively in your
environment.