Elden Christensen | Principal Program Manager Lead | Microsoft

Symon Perriman | Vice President | 5nine Software

Meet Elden Christensen
• Elden is a Principal Program Manager Lead on the
Microsoft Windows Server high availability product
team.

• A high availability specialist, Elden has been with Microsoft since

1999 and is responsible for the Failover Clustering and Network
Load Balancing features in Windows Server.
Meet Symon Perriman | @SymonPerriman

• VP of Business Development & Marketing for 5nine Software,

a Hyper-V security & management provider
• Previously Microsoft's Senior Technical Evangelist for Windows Server &
System Center and a Program Manager for Failover Clustering
• Holds several patents and co-authored "Introduction to System Center
2012 R2 for IT Professionals" (Microsoft Press)
• Graduated from Duke University with degrees in Computer Science,
Economics and Film & Digital Studies
• Technical advisor for several startups, including System Center solutions
provider ScorchCenter
Course Modules

Failover Clustering in Windows Server 2012 R2

01 | Introduction to Clustering 05 | Hyper-V Clustering

06 | Multi-Site Clustering &

02 | Cluster Deployment & Upgrades
Disaster Recovery

07 | Advanced Cluster Administration &

03 | Cluster Networking
Troubleshooting

08 | Managing Clusters with

04 | Cluster Storage & Scale-Out File Server
System Center 2012 R2
Setting Expectations

• Target Audience
– IT Professionals interested in learning about high-availability (HA)
– Hyper-V Virtualization administrators looking for VM HA
• Suggested Prerequisites
– A basic understanding of storage, networking and virtualization
– Some experience with Windows Server 2012 R2
– No prior clustering knowledge is needed
– Course will start at L200, mostly run at L300, and end at L400
• Clustering with SQL and Exchange will not be covered
Recommended Resources

• Cluster Team Blog

http://blogs.msdn.com/b/clustering/

• Clustering on TechNet
http://technet.microsoft.com/en-us/library/hh831579.aspx

• Clustering Forum
– https://social.technet.microsoft.com/Forums/windowsserver/en-
US/home?forum=winserverClustering

• Clustering Virtual Labs:

– Windows Server 2012 R2: http://go.microsoft.com/?linkid=9846211
– System Center 2012 R2 Virtual Machine Manager:
http://go.microsoft.com/?linkid=9846205
Elden Christensen | Principal Program Manager Lead | Microsoft
Symon Perriman | Vice President | 5nine Software
Module Overview

• Introduction to Failover Clustering

• Hardware High-Availability
• Workload High-Availability
• Storage High-Availability
• Guest Clustering
• Multi-Site Clustering
• Management High-Availability
Introduction to Failover Clustering
What is High Availability?

• The world is now a 24/7 global marketplace

• Systems must be online or customers are lost
• Goal of high availability (HA) is to keep systems, applications,
services, email, databases, files & printers always running
• Every business now has high availability needs
Uptime Percentage (%) Downtime per year
99.999 (“Five Nines”) 5 minutes
99.99 52 minutes
99.9 8.7 hours
99 3.7 days
Why is HA Important?

• Server downtime is inevitable

• Planned downtime
– Maintenance
– Upgrade
• Software or Hardware
– Update
• Hotfix, Security Patch
• Unplanned downtime
– Disaster
– Power Outage
– Accident
• Start planning now!
Network Load Balancing

• Distributes network traffic across servers to balance load

• Ideal for stateless servers with redundant or identical data
• Appears like a single, scalable and highly-available server
• Load-balance across Hyper-V VMs NLB Cluster
Failover Clustering 101

1. Deploy redundant hardware everywhere

2. “Shared” storage accessible by all nodes
3. Clustered application writes data to shared storage so all
nodes can assess it
4. Registry settings containing current cluster configuration
information is replicated to each node
5. Nodes monitor the health of other nodes
Failover Clustering 101

6. If a node fails, the health check will fail,

and a failover will happen
7. Another node knows what the failed node was hosting
through the registry and start that application
8. The application reads its saved data from
the shared storage
9. Client connects to app on new node
10. Clients may experience a slight interruption in service during
failover but none during a live migration (of a VM)
Overview of a Failover

Public
VMs & VMs &
Workloads Workloads

Shared
Storage
High Availability throughout the Datacenter
1. Hardware High Availability
– Servers, storage, networking, etc.

2. Workload High Availability

– Host Failover Clustering for VMs & Roles

3. Guest Application High Availability

– Guest Failover Clustering for apps within VMs

4. VM Storage High Availability

– Scale Out File Server Failover Clustering

5. Management High Availability

– System Center

6. Site High Availability

– Hyper-V Replica & Multi-Site Clustering
Hardware High-Availability
Hardware High Availability
• Redundancy everywhere
• Server
– Redundant server roles (AD, DHCP, DNS, etc.),
– System: Hot swapping, BMC sensors, Power protection
– Processor: Instruction error detection, instruction retry, lock-step
processors, machine check architecture, extended precision
– Memory: Windows Hardware Error Architecture (WHEA), parity bits, error
correcting code, memory scrubbing, bad page offloading

• Storage
– Multi-Path IO (MPIO), RAID, checksums, background scrubbing, resilient
file systems

• Networking
– Multiple networks, NIC Teaming, Load Balancing (NLB), Multi-Channel SMB
Failover Clustering Scale

• Increased scale out and scale up

– 4x scale over Windows Server 2008 R2 ...

– 64-nodes in a cluster
– 8,000 VMs in a cluster

Scale up
– 1,024 VMs per node
• Microsoft Hyper-V Server 2012 R2
– Same clustering & Hyper-V features
...
– Live migration, CSV, 64-nodes, etc.
– FREE Scale out
Failover Cluster Support Policy

• Servers: Two simple requirements for Cluster support

1. All components must have a logo

http://server.windowsmarketplace.com/Server
• Storage (SAN, iSCSI Target)
• Network Interface Card (NIC)
• Host Bus Adapter (HBA)
• * NIC Teaming Software
• * MPIO/DSM (Multipathing Software)

2. Complete solution must pass Cluster Validation

http://technet.microsoft.com/en-us/library/cc732035.aspx
Cluster Storage Options
Flexible storage choices for building clusters

Shared Storage Data Replication

FC SAS RBOD Spaces RAID HBA SMB Hardware Software Application

Replication Replication Replication
SAS JBOD

3rd party Example:

software Exchange
iSCSI FCoE replication SQL AlwaysOn
solution
Networking
• Integrated with Windows Server TCP/IP Stack
• Networks configurable by priority & function
• Multiple networks for added redundancy
– Public Network
– Internal Cluster Network
– Hyper-V Live Migration & Management

• IPv4, IPv6 & IPv6 Tunneling

• DHCP or Static IP Addresses
• NIC Teaming supported on all networks
• Nodes can reside in different subnets, used in multi-site clustering
Workload High Availability
Workload High Availability
• Physical servers create a failover cluster
• Survive Host Crashes
• VMs restarted on another node

• Restart VM Crashes
• VM OS restarted on same node

• Recover VM Hangs
• VM OS restarted on same node

• Zero Downtime Maintenance & Patching

• Live migrate VMs to other hosts

• Mobility & Load Distribution

• Live migrate VMs to different servers to load balance
HA Roles and Features

• Common • Other
– DFS-Namespace
– Hyper-V – DFS-Replication
DHCP
– Exchange –
– DTC
– File Server – Hyper-V Replica Broker
– IIS
– SQL – iSCSI Target
– iSNS
– MSMQ
• Generic Containers – NFS
– Remote Desktop
– Generic Application – WINS
– Generic Script
• 3rd Party
– Generic Service
Storage High Availability
VM Storage High Availability
• Scale Out File Server
• Virtual hard disk (VHD)
on a file share
SMB Client

• Multiple client access

points (\\SOSF) \\SOFS \\SOFS

• Resilient during
failover and crash
Share1 Share2 Share1 Share2

Node 1 Node 2

Cluster
Guest Clustering
Guest Application High Availability
• VMs create a (virtualized) failover cluster
• Guest Application
Health Monitoring
– Application restarts or fails over
– Detect blue screens & user mode hangs
– VM network availability

• Application Mobility
– Guest OS needs patching or VM needs
maintenance, application moved to
the other node
Multi-Site Clustering
Site High Availability

• Nodes in different physical locations

• Survive the loss of an entire datacenter
• Stretch sites over a large distance
• Storage at both sites with replication
• Automatic or manual
recovery
• Technologies
– Multi-Site Clustering
– Hyper-V Replica
– Azure Site Recovery
Management High Availability
 Management High Availability
• Server
– Redundant server deployments
– Run server inside a clustered VM
– Backup using DPM or Replicate using
Hyper-V Replica
– Monitor with a SCOM Management Pack

• Database
– SQL Server 2012 AlwaysOn Clustering
– Replication / Mirroring / Backup to a
secondary site
– Run SQL inside a clustered VM
– Backup using DPM or Replicate using
Hyper-V Replica
– Monitor with a SCOM Management Pack
©2015 Microsoft Corporation. All rights reserved. Microsoft, Windows, Office, Azure, System Center, Dynamics and other product names are or may be registered trademarks and/or trademarks in the
U.S. and/or other countries. The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft
must respond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after
the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.