Setting up for client-side data deduplication

Edit online

You must edit the client options file before Data Protection for Oracle can use client-side data deduplication through the IBM Storage Protect™ API.

About this task

You can turn on client-side data deduplication by adding DEDUPLICATION YES to the dsm.sys file and by making sure that the deduplication prerequisites are met.

You can turn on client-side data deduplication by adding DEDUPLICATION YES to the dsm.opt file and by making sure that the deduplication prerequisites are met.

The IBM® Storage Protect server administrator must enable data deduplication for the Data Protection for Oracle with the appropriate server command. For example:

UPDATE NODE ORACLE_NODE DEDUPLICATION=CLIENTORSERVER

The IBM Storage Protect server administrator must enable data deduplication on the storage pool where the Oracle data is stored with the following server command:

UPDATE STGPOOL BACKUP_POOL DEDUPLICATION=YES

Results

After you created backups with client-side data deduplication enabled, you can use the tdposync query command to verify that client deduplication occurred during the backup operation. For detailed statistics, you can also query the IBM Storage Protect server activity log for the total data reduction.

You can also use the performance monitor feature in the IBM Storage Protect server to verify the percentage of data that has been deduplicated. The performance monitor feature is part of the IBM Storage Protect Administration Center. The data deduplication statistics are displayed graphically in the Performance GUI in the Administration Center.

The following example illustrates how you can set up the dsm.sys file on AIX® to enable the performance monitor feature:
servername fvtseries2
tcps fvtseries11esx2.storage.usca.ibm.com
tcpp 1500
nodename apitest
*errorlogname /home/api/logs/tdperrs.log
errorlogname  /home/orc11r2/tdperrs.log
PERFMONTCPSERVERADDRESS jumboesx1.storage.usca.ibm.com
PERFMONTCPPORT 5129
Considerations:
  • The deduplication and enablelanfree options are mutually exclusive. Therefore, you can use either one option or the other, but not both options together.
  • The deduplication and enableclientencryptkey options are also mutually exclusive. Therefore, you can use either one option or the other, but not both options together.
  • A local deduplication cache is an optimization that can reduce network traffic between the IBM Storage Protect server and the client. Client-side data deduplication can occur with or without it. Do not use the deduplication cache with Data Protection for Oracle for the following reasons:
    • The cache cannot be used when multiple processes, such as concurrent backups or IBM Storage Protect API applications, transfer content concurrently. Data Protection for Oracle backup operations that use multiple channels use multiple processes.
    • It is possible that the client deduplication cache can become out of sync with the server-deduplicated disk storage pool. This state can be the result of object expiration, file space deletion, and overflow to an associated tape storage pool. When the client cache contains entries that are no longer in the IBM Storage Protect server deduplicated pool, the cache is reset and the backup operations fails. The IBM Storage Protect API does not attempt the backup again.
  • When IBM Storage Protect server expiration or a similar process that removes deduplicated data extents runs concurrently with a deduplicated backup, the backup might fail. Backup operations with client-side deduplication enabled fails with the following messages:
    • Return code=254
    • Error message: ANS7899E The client referenced a deduplicated extent that does not exist on the TSM server.