Sie sind auf Seite 1von 45

Using TSM Active Data Pools for On-Site and Off-Site Backup and Restore

Dave Daun IBM Advanced Technical Support djdaun@us.ibm.com

2009 IBM Corporation

IBM Advanced Technical Support

Agenda
Overview of Active Data Pools Defining Active Data Pools Removing Inactive Files from an Active Data Pool Restoring from Active Data Pools Restrictions and Potential Issues Active Data Pools Offsite Volumes Implementation Ideas Demonstration
2
2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools


New feature in TSM 5.4
Long standing requirement from the TSM community New Storage Pool Type Only for Backup data Only for Sequential Access pools Implementation is similar to Copy Storage Pools Restore will come automatically from the Active Data Pool

Why use Active Data Pools


Faster restore of data Useful for disaster recovery Minimizes tape mount activity during restore

New function in TSM V6 Offsite functions using Disaster Recovery Manager (DRM)
3
2009 IBM Corporation

IBM Advanced Technical Support

TSM Pool Types

Primary

Random Disk

Sequential Tape

Sequential Disk (FILE)

Copy Pool

Sequential Disk Sequential Tape (FILE)

Active Data Pool

Sequential Tape

Sequential Disk (FILE)

2009 IBM Corporation

IBM Advanced Technical Support

Collocation of Active Data


New
Storage hierarchy contains active and inactive data Active-data pool on sequential-access disk for fast restore

Active/inactive data in copy pool for disaster recovery

Active backup versions are those that still reside on client machine (have not been modified or deleted)

Active data pools are sequential-access--typically sequential-access disk. Can also be on tape.
Active backup versions exist in both a primary pool and one or more active data pools. Active data pools are populated either synchronously with backup or asynchronously (similar to copy pools)
Synchronous copy via a primary pool definition Asynchronous copy with new command: COPY ACTIVEDATA Asynchronous only for LAN-Free backups 5
2009 IBM Corporation

IBM Advanced Technical Support

Removing Inactive Versions from Active Data Pools


Inactive versions in a Active Data Pool are treated as expired files
Version is inactivated when a new backup version is received from the client Reclamation removes inactive versions from Active Data Pools Reclamation must be run regularly MOVE DATA or MOVE NODEDATA will also remove inactive versions from an Active Data Pool
RECONSTRUCT defaults to YES on MOVE DATA or MOVE NODEDATA in an Active Data Pool

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Disk vs. Tape


Benefits of Active Data Pools on sequential disk
Optimized access to active versions for fast restore Reduced size of disk pool if only active versions are stored Reduced data movement in preparation for restore of active data

Benefits of Active Data Pools on sequential tape


Reduced storage requirement (onsite or offsite) while protecting data against media failure or disaster Simplified tape management because of fewer tapes
7
2009 IBM Corporation

IBM Advanced Technical Support

Defining an Active Storage Pool


Define the pool DEFINE STGPOOL ACTIVE_POOL POOLTYPE=ACTIVEDATA

Add the active data pool to an existing primary pool definition (optional)
UPDATE STGPOOL PRIM_POOL ACTIVEDATAPOOLS=ACTIVE_POOL

Schedule a command to copy from the primary pool to the active pool
COPY ACTIVEDATA PRIM_POOL ACTIVE_POOL MAXPROC=4 Create a new policy domain to enable active data pools (or update an existing domain) DEFINE DOMAIN ACT_DOMAIN ACTIVEDESTINATION=ACTIVE_POOL
8
2009 IBM Corporation

IBM Advanced Technical Support

Controlling Which Nodes Use Active Data Pools


Policy Domain ACTIVEDESTINATION Active Pool Policy Set

Management Class

Primary Pool NODES


Copygroup

2009 IBM Corporation

IBM Advanced Technical Support

Controlling Which Nodes Use Active Data Pools


1. Use an existing domain: Simply use the UPDATE DOMAIN command and specify the list of active-data pools in the ACTIVEDESTINATION parameter.

2. Restrict which nodes can write to an active-data pool:

Decide which nodes are critical and should use Active Data Pools

Define a new domain and specify the list of active-data pools in the
ACTIVEDESTINATION parameter. Update the node definitions to assign the nodes to the new domain Complete the policy structure (copy it if possible)

10

2009 IBM Corporation

IBM Advanced Technical Support

Getting Active Versions Into an Active Data Pool


Using simultaneous copy (not available for LAN-Free) New parameter on DEFINE and UPDATE STGPOOL command ACTIVEDATAPOOLS Specifies the names of active-data pools where the server simultaneously writes data during a client backup operation. The combined total number of storage pools in the COPYSTGPOOLS and the ACTIVEDATAPOOLS cannot exceed three. COPY ACTIVEDATA command Use this command to copy active versions of backup data from a primary storage pool to an active-data pool. Similar to BACKUP STGPOOL Should be used in addition to simultaneous copy
11
2009 IBM Corporation

IBM Advanced Technical Support

Removing Inactive Files from an Active Data Pool


When a file is deactivated, the logical occupancy of the volume it is stored on is reduced by the size of the file. This causes reclamation to run sooner for the volume.
Reclamation (with aggregate reconstruction) is used to remove inactive files from the active data pool. Reconstruction will build the new aggregate by copying just the active files and not any deleted or inactive files. The COPY ACTIVEDATA command and MOVE DATA/NODEDATA commands will reconstruct aggregates as they are copied or moved within an active-data pool. Do not specify RECONSTRUCT=NO on the MOVE DATA/NODEDATA commands when dealing with activedata pools.

12

2009 IBM Corporation

IBM Advanced Technical Support

Collocation of Active Data An Example


1. Client backs up A0, B0, C0, D0 to primary pool with simultaneous write to active data pool.
A0 B0 C0 D0

2. Client backs up B1, E1 with simultaneous write to active data pool. B0 deactivated.
B1 E1

Client Active data pool Active/inactive primary pool

Server
A0 B0 C0 D0 A0 B0 C0 D0

Client Active data pool Active/inactive primary pool

Server
A0 B0 C0 D0 B1 E1 A0 B0 C0 D0 B1 E1

3. Reclamation removes inactive B0 from active data pool.

4. Client restores active files A0, C0, D0, B1, and E1 from active data pool.
A0 C0 D0 B1 E1

Client Active data pool Active/inactive primary pool

Server
A0 C0 C0 D0 B1 E1 B0 D0 B1 E1 A0 B0 C0 D0 B1 E1

Client Active data pool Active/inactive primary pool

Server
A0 C0 C0 D0 B1 E1 B0 D0 B1 E1 A0 B0 C0 D0 B1 E1

13

2009 IBM Corporation

IBM Advanced Technical Support

Client data restore when using Active Data Pools


When restoring client data, the server will select the active version of a file from an Active Data Poll when appropriate. The restore order has been modified to include Active Data Pools:
1. Active Data Pool (FILE) 2. DISK (random)

3. FILE (primary or copypool)


4. Active Data Pool sequential tape onsite. 5. Sequential onsite-volume (primary, copy or Active Data Pool), priority considers: mounted(idle) -> automated -> manual
14
2009 IBM Corporation

IBM Advanced Technical Support

Restoring Volumes or Storage Pools with Active Data Pools


Two new parameters on the RESTORE STGPOOL and RESTORE VOLUME commands:
ACTIVEDATAONLY
Specifies that active versions of backup files are to be restored from active-data pools only.

ACTIVEDATAPOOL
Specifies the name of the active data pool from which the active versions of backup files are to be restored. If this parameter is not specified, files are restored from any active data pool in which active versions of backup files can be located.

Note: Inactive files cannot be restored from active data pool and will be deleted from the primary pool (when the volume is deleted)
15
2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools - The Q OCC and Q CONTENT commands


Query Occupancy Command: Number of Files

Reduced when non-aggregated files are deleted during deactivation Reduced by number of logical files only when aggregate is deleted

Physical Size reflects the actual size of the aggregates (as it does today) Logical Space Occupied
Reduced by size of logical file being deleted Reduced by size of logical file that is removed during reclamation (reconstruction)

Query Content Command:


Will show all logical files in an aggregate regardless if it has been removed
16
2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Restrictions and Potential Issues


Can only be used for Backup data, not Archive data or HSM data
NDMP Backups do not go to Active Data Pools A point-in-time restore will not use Active Data Pools Must plan for time needed to do active data copies Concurrent copies may be slower COPY ACTIVEDATA may be an additional process COPY ACTIVEDATA runs somewhat slower than BACKUP STGPOOL Difference is due to additional processing needed to verify the active/inactive state of logical files in aggregates and act accordingly. Must check ownership of files and see of Node is authorized to used Active Data Pools for each file
17
2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools for Off-Site


Active Data Pool volumes will be processed by the PREPARE command and by the Query DRMedia and MOVE DRMedia commands
Active Data Pool volumes can have an access mode of OFFSITE Active Data Pool volumes with an access mode of OFFSITE will use the off-site reclamation process (data is read from Primary Pool volumes) The implementation is nearly identical to Copy Storage Pools

18

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools for Off-Site - Command Changes


SET DRMACTIVEDATASTGPOOL New command
QUERY DRMSTATUS Output now shows Active Pools UPDATE VOLUME - no change OFFSITE for ADP volumes

MOVE DRMEDIA new parm: ACTIVEDATASTGPOOL=


PREPARE new parm: ACTIVEDATASTGPOOL= QUERY DRMEDIA new parm: ACTIVEDATASTGPOOL=

19

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pool Implementation Ideas


Use an Active Data Pool as a substitute for disk only pool
Reduces the size of the disk requirement

Faster restores from disk (non-point-in-time restores)


Backups Primary Pool Active Pool

Migration Tape Copy Pool (for offsite)

20

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pool Implementation Ideas


Use an Active Data Pool for on-site media recovery
Reduces the size of the tape requirement

Faster restores from tape (non-point-in-time restores)

Backups

Primary Pool Active Pool

Migration Tape Copy Pool (for offsite)

21

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pool Implementation Ideas


Use an Active Data Pool as a staging area for an anticipated restore
Reduces the size of the disk staging area Faster restores when enough time is available to perform the staging operation Use a special Policy Domain to enable a single node copy
Backups

Primary Pool
Active Pool Copy when needed Migration Tape Copy Pool (for offsite)

22

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pool Implementation Ideas


Use an Active Data Pool as a staging area for recurring Backupset generation
Reduces the number of tape mounts needed for Backupset generation
Use Policy Domain to restrict Active Pool to only those nodes that need Backupsets
Backups Primary Pool Active Pool Backupsets

Migration Tape Copy Pool (for offsite)

23

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pool Implementation Ideas


Use an Active Data Pool for off-site vaulting
Reduces the size of the off-site tape requirement

Faster disaster restores from tape (very few inactive files on the disaster tapes)
Use DRM to manage off-site Active Data Pool volumes
Backups Primary Pool Active Pool (for off-site) Migration

24

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pool Implementation Ideas


Use an Active Data Pool as an electronic vaulting target

Reduces the amount of data which must be stored in the remote location
Faster disaster restores

Replaces the Copy Storage Pool


Use DRM to manage the off-site volumes
Backups
Primary Pool Extended SAN Active Pool

Migration

25

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pool Demonstration Server Layout


NODE DEVCLASS STGPOOL VOLUMES POOL101 POOL102

DISKPOOL1 Random Primary Pool


DISK ACTNODE1 (ACTDOM1) DISKPOOL2 Random Primary Pool

POOL201

ACTNODE2 (ACTDOM2)

ACTDEV

ACTPOOL Sequential Active Data Pool

VOL001 VOL002 .. VOL014

26

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pool Demonstration


Server Scripts
FILEVOLS displays the status, utilization and percent reclaimable of the ACTPOOL volumes RECLAIM runs reclamation on the ACTPOOL with a threshold of 10%

ACTLOG5 displays the last five minutes of Activity Log records


Windows Scripts Touchfiles updates the last access date of specified number of files

27

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pool Demonstration


Highlights
Backup repeatedly and run reclamation to see the inactive files removed from the active pool Backup with concurrent copy to the active pool Use the COPY ACTIVEDATA command to populate the active pool Restore data from an active data pool Perform off-site storage pool reclamation

28

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Q DOMAIN

29

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Q STGPOOL

30

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Q STGPOOL

31

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools COPY ACTIVEDATA

32

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools The FILEVOLS Script

33

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Day 1: First Incremental

34

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Day 1: First Incremental

35

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Day 2: Subsequent Incremental

36

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Day 2: Subsequent Incremental

37

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Day 2: Reclamation

38

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Day 2: After Reclamation

39

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Day 3: After Reclamation

40

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Client Restore

41

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Actlog for Client Restore

42

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Offsite Reclamation

43

2009 IBM Corporation

IBM Advanced Technical Support

Active Data Pools Offsite Reclamation

44

2009 IBM Corporation

IBM Advanced Technical Support

Questions???

45

2009 IBM Corporation

Das könnte Ihnen auch gefallen