Home » RDBMS Server » Backup & Recovery » Backup data over 200GB while online (Oracle 10.2.0.4, Red Hat Linux 5.3)
Backup data over 200GB while online [message #414285] Tue, 21 July 2009 06:33 Go to next message
linuxos
Messages: 31
Registered: April 2009
Member
I don't have much knowledge in backup and Oracle. I found that the performance is not good with RMAN backup, and at the same time affecting some online processing.

I'm thinking whether it's normal to specify "RATE=10M" while I've >200GB data, so that the bandwidth will not be fully occupied by RMAN? How can I know the optimal RATE?

I also want to use 2 channels in order to speed up the backup. How do I know the resource used by one channel? Will it use memory from SGA or from operating system?

configure device type disk parallelism 2 backup type to compressed backupset;


Thank you very much!
Re: Backup data over 200GB while online [message #414294 is a reply to message #414285] Tue, 21 July 2009 07:23 Go to previous messageGo to next message
ebrian
Messages: 2794
Registered: April 2006
Senior Member
linuxos wrote on Tue, 21 July 2009 07:33
I found that the performance is not good with RMAN backup, and at the same time affecting some online processing.

What makes you say this? Are you performing your backups at off hours?

Depending on what your performance issue is, you can consider MINIMIZE LOAD, RATE and/or implementing a change tracking file.

Increasing the parallelism "could" have a negative impact on performance.

The first thing is determining where the "slowness" is.
Re: Backup data over 200GB while online [message #414530 is a reply to message #414285] Wed, 22 July 2009 08:04 Go to previous message
linuxos
Messages: 31
Registered: April 2009
Member
I tried the parallelism, it is quite fast (but there is no data update during the period). Besides, setting a maxpiecesize to a proper value may also help performance. Cool
Previous Topic: Incremental backup not marking obsolete files
Next Topic: RMAN backup used cross platform
Goto Forum:
  


Current Time: Sun Dec 11 04:39:38 CST 2016

Total time taken to generate the page: 0.14756 seconds