Re: RMAN backup optimization

Hello Suresh

Maybe you should try some of the following and see if it helps :

1. Suggested workarounds for Bug 6476935<https://metalink.oracle.com/metalink/plsql/showdoc?db=Bug&id=6476935> (metalink node *463227.1)*
2. Taking backups without compression (If disk space is not an issue).
3. Review database performance report during the times when performance is a bottleneck.
4. Verify that there are enough file descriptors /max file limit at the O.S level.

-Rajeev

On Dec 26, 2007 12:57 PM, Suresh Chaganti <chaganti.suresh_at_gmail.com> wrote:

> Hello List
>
> I just got onboard at this client and looking into some performance
> issues.
>
> HW / SW : 2 Node RAC, 10.2.0.3, 32 CPU, ASM, 32 GB RAM ~ 1.5 TB database
>
> The redolog file size is 2GB and about 20GB redo is generated every hour.
>
> The archive logs are on one of the disk groups of ASM and there is a RMAN
> job running every 2 hrs backing up archivelogs.
>
> Typically, this job is runnning for 90 minutes to backup about 40GB of
> archive redo.
>
> There is concern that this RMAN job is hogging too many resources and
> impacting production application performance. I am still in process of
> getting familiar with application and environment.
>
> Here is the rman script
>
> configure backup optimization on;
> run {
> allocate channel d1 type disk maxpiecesize 1024M maxopenfiles ${FP};
> allocate channel d2 type disk maxpiecesize 1024M maxopenfiles ${FP};
> allocate channel d3 type disk maxpiecesize 1024M maxopenfiles ${FP};
> allocate channel d4 type disk maxpiecesize 1024M maxopenfiles ${FP};
> allocate channel d5 type disk maxpiecesize 1024M maxopenfiles ${FP};
> allocate channel d6 type disk maxpiecesize 1024M maxopenfiles ${FP};
> backup as compressed backupset tag PROD archivelog all format
> '/u02/RMAN/arc/al_%d_%U.rman';
> release channel d1;
> release channel d2;
> release channel d3;
> release channel d4;
> release channel d5;
> release channel d6;
> }
> configure backup optimization off;
> EOF
>
> Is 90 Minutes to backup 40GB of archived redo on this box sound normal?
>
> The backup was running on 2 channels, but was recently modified to 6
> channels and users starteed complaining about reduced performance.
>
> One idea came to my mind is investigate if we are better off moving
> archive destination outside ASM on OCFS or any cluster FS and use OS
> utilities to compress and backup. Any idea if this will be faster than RMAN
> backup
>
> Thanks
> Suresh
>

--
http://www.freelists.org/webpage/oracle-l