Oracle backups to NetBackup Deduplication Appliance 5020 perform slowly, around 1 MB/second or less.

Article: 100027321
Last Published: 2013-10-28
Ratings: 0 0
Product(s): NetBackup & Alta Data Protection

Problem

Oracle backups to NetBackup Deduplication Appliance 5020 perform slowly, around 1 MB/second or less.

Cause

Default size data buffer, number data buffers.

Solution

On each NetBackup media server performing RMAN backups of Oracle to deduplication storage:
1. create/edit .../netbackup/db/config/SIZE_DATA_BUFFERS_DISK w/value 1048576
2. create/edit ../netbackup/db/config/NUMBER_DATA_BUFFERS_DISK w/value 256

Note: No restart is needed for NetBackup after these files are created.
NOTE2: If Status 89 is now seen on the backups, increase the system's shared memory value in the kernel settings and then reboot the system. For a workaround on the Status 89, temporarily reduce the values above in the buffer files and retry the backup.

3. In the Oracle client RMAN  stream-based backup script:
    a. Increase the # of allocated channels above the current four to 6 then test, 8 then test, 10 then test, 12 then test, etc.

If stream-based RMAN  backups are not desired, one can also try RMAN  proxy copy backup method. examples can be found on bottom of page 147 through  the top of page 150 in the  Veritas NetBackup™ for Oracle Administrator's Guide UNIX, Windows, and Linux Release 7.1 and on pages 16-17 of the NetBackup Deduplication Guide (linked in Related Articles)

IMPORTANT:  After making any changes, it is vitally important to verify that the following tests can be completed successfully:
1. Run a backup
2. Restore data from the new backup created after changing SIZE_DATA_BUFFERS and NUMBER_DATA_BUFFERS
3. Restore data from a backup created prior to the changes to SIZE_DATA_BUFFERS and NUMBER_DATA_BUFFERS

 

Applies To

- NetBackup: master version 7.1.0.3 on Solaris SPARC / Media server 7.1.0.3 on 64bit Red Hat Enterprise Linux 5, Oracle clients are remote.

- PureDisk:
2 nodes in each storage pool
2 pools replicating to each other using SLP's to two PDDO pools
All 5020's running 1.4.1.1+ ET2701800 v5 + EEB from ET2669584 (MBRTws.php)

- clients communicate over 1GB Ethernet to mediasvr same subnet while media server communicates over 10GB to PureDisk appliances.


set echo off;
@/vol/dbabin/rman/connect_rman
set echo on;

run
{
allocate channel c1 type 'SBT_TAPE';
allocate channel c2 type 'SBT_TAPE';
allocate channel c3 type 'SBT_TAPE';
allocate channel c4 type 'SBT_TAPE';
send ' NB_ORA_POLICY=db_backup,NB_ORA_SCHED=ApplicationBackup_sched';
backup incremental level = 0 filesperset=1 cumulative database include current controlfile plus archivelog ;
}

Was this content helpful?