Video Screencast Help
Symantec to Separate Into Two Focused, Industry-Leading Technology Companies. Learn more.

RMAN backup/restore using DDBoost

Created: 19 Jun 2013 | 6 comments

We moved RMAN Oracle backups from DD VTL (on NB 7.0) to OST (on NB7.5) recently.  DBA was running a db restore and wants to know what we can do to speed up the restore since he says it is much slower now.  He is allocating 3 channels (we have max. jobs per client set to 3).  He was restoring a 300G db that took a little over an hour. We have Linux master and media servers. Thanks.

Operating Systems:

Comments 6 CommentsJump to latest comment

StefanosM's picture

Deduplication is good for many thinks (every deduplication, netbackup, dd, storeone) but the restore is not one of them.

I do not know match of DD but I suspect that the backup stream is rehydrated at the DD appliance. So the hole backup stream comes from the DD.
You can not compare the speed of a VTL with the speed of the network. At least if you have 1 Gbps cards and/or the DD is not powered enough.

Try to use more channels (5, 10 or more ) and I suppose that you have read the DD best practices for oracle (and netbackup deduplication best practices for oracle is essentially the same document)

huanglao2002's picture

1 we want to know restore speed before upgrade to 7.5, do you have some base line info?

2 from you post,you only tell use upgrade netbackup software, Do you have alse upgrade DD OST plug-in?

3 please check the OST plug-in and netbackup SCL.

Nicolai's picture

I have never heard of I/O issues on a Data Domain. I have both DD890 and DD990 and they restore without any issues 800MB/sec using 4-5 stream combined.

I suspect the issue is related to  :

  • Network limitation (both client/media server/DD)
  • I/O limitation on the client (receiving 300GB in a hours is almost 100MB/sec).

If you have Unix media server or Unix clients you can easily verify operation state using the GEN_DATA file directive.

http://www.symantec.com/docs/TECH75213

Assumption is the mother of all mess ups.

If this post answered your'e qustion -  Please mark as a soloution.

Kim Kelly's picture

Thanks for all the replies. I will have him allocate a higher number of channels (the client max jobs is actually set to 12) and see if we can bump up the throughput.  It may be just his perception that the restore is slower.  We do not have any benchmarks from the 7.0 environment.

We have DD990 running 5.2.2.0 with the 7.5 compatible Boost plugin 2.5.1.0 on all the media servers.

Nicolai's picture

I would be very gentle putting my faith on gut feeling without actual numbers.

One hint - if you have 10G Ethernet infrastructure ensure all optical cables are clean before they are inserted into SFP's. 10G is very sensitive to contamination

Assumption is the mother of all mess ups.

If this post answered your'e qustion -  Please mark as a soloution.

rookie11's picture

As per your configuration:

We have DD990 running 5.2.2.0 with the 7.5 compatible Boost plugin 2.5.1.0 on all the media servers.

use 4 channels in RMAN backup to get backup done in minimum time. same number of channels can be used to do restore. i have tested it my enviorment on both test and production databases.