Networker

Re: [Networker] Making NDMP selective backups faster

2007-04-05 11:27:29
Subject: Re: [Networker] Making NDMP selective backups faster
From: Stan Horwitz <stan AT TEMPLE DOT EDU>
To: NETWORKER AT LISTSERV.TEMPLE DOT EDU
Date: Thu, 5 Apr 2007 11:21:51 -0400
On Apr 5, 2007, at 9:48 AM, Fazil Saiyed wrote:

Hello,
Could you explain a bit the selective backup method you are employing, to my knowlede, NDMP backups are block level with capability to restore at
file level ( db file in your case), i am not aware of any modules that
lets you backup or restore to file level with NDMP backups withen a dB.

I am using the client resource options "type=image, hist=y, update=n" in the client resource. The backup level is controlled by the client's group, and it is set to full.

You mentioned the GB links, so if you are using over the network method, NDMP funtionality of block level backup does not change, however, you may
want to attach the drive directly to NDMP mirapoint server to use FC
speeds, usually, in most cases attaining native tape drive speed, this
will help your backups a lot, if i saw your logs correctly, it took you over 25 hrs to backup just close to 600 GB,in my experiance with good NAS
performance ( NDMP-V4) in my case that much data can be backed up with
LTO2 in close to 6 hrs ( Netapp FAS3050c) in ndmp method.

Attaching the tape drives to our Mirapoint servers would be prohibitively expensive. According to the Mirapoint engineers with whom I have consulted, doing so would only give us a modest boost in throughput.

With the options (type=image, hist=n, update=n), I typically get around 13 hours, but these are email servers and they are almost always heavily loaded by users who access their email via the box's webmail software, IMAP, or POP.

So unless the mirapoint is performing poorly, you should attain heiger
speeds.

One would think, except that I have never seen more than 11MB/s going to our tape drives, even from a lightly loaded server that we use only for R&D purposes.

Adding storage node is a good idea in your case where lot of your backups are over the network to the MS, the reason you did not see improvments is probably because the server had different bottlenecks, i,e Network or your
clients and not the server itself.

Try using some testing with native ndmp dump to the tapes, bigasm, or MT
to verify your tape setup.

Unfortunately, doing testing other than NetWorker server-initiated NDMP backups is my only option. Other than the variables I mentioned previously, I have no control over how the backups work from the Mirapoint side because it is a heavily closed system.

As far as server size, buy something that is reasonably fast with room to
grow if bottlenecks appear.

Thanks

To sign off this list, send email to listserv AT listserv.temple DOT edu and type 
"signoff networker" in the body of the email. Please write to networker-request 
AT listserv.temple DOT edu if you have any problems with this list. You can access the 
archives at http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

<Prev in Thread] Current Thread [Next in Thread>