Veritas-bu

Re: [Veritas-bu] Catalog performence to tape

2007-07-10 14:43:27
Subject: Re: [Veritas-bu] Catalog performence to tape
From: Joe Royer <jroyer AT digitalmotorworks DOT com>
To: "Clooney, David" <david.clooney AT bankofamerica DOT com>
Date: Tue, 10 Jul 2007 13:32:04 -0500 (CDT)
Admittedly, flushing the small DSSU on the same spindles is more than 
twice as fast as the catalog.  As I have learned from this thread, NBU 
does not honor the tuned data buffer size when doing a catalog backup, 
but I'm skeptical it could make that huge of a difference.  64k vs 32k

Oh, and my example is NBU 5.1 MP6 on linux.


On Tue, 10 Jul 2007, Clooney, David wrote:

> One question would be is , how does the backup to tape work at
> acceptable speeds when using a NBU policy in the same environment ?
>
> Dave
>
> -----Original Message-----
> From: veritas-bu-bounces AT mailman.eng.auburn DOT edu
> [mailto:veritas-bu-bounces AT mailman.eng.auburn DOT edu] On Behalf Of Joe
> Royer
> Sent: 10 July 2007 15:49
> To: veritas-bu AT mailman.eng.auburn DOT edu
> Subject: Re: [Veritas-bu] Catalog performence to tape
>
> I have gone a few rounds with this issue myself, and here is what I
> learned:
> - Catalog disk needs read performance, not write.
> - RAID-5 is faster than RAID-10 for sequential reads.
> - If you put your catalog on SAN disk, don't use the same HBA/port that
> you use for tape.
> - Most servers' internal RAID controllers are brain dead
>
> My catalog is about 70GB and I can back it up from internal disk
> to LTO2 in about 50 minutes.  Given what I have seen in the past, I
> can't
> complain about this number.
>
> My master is an IBM x345 with six U320 SCSI drives.  Once upon a time
> (before me), it was configured with all six drives in a single RAID-5
> and
> when the catalog ran the OS would suffer because it was all the same
> virtual disk.  I had an opportunity to rebuild the box and I thought I'd
>
> be smart and change it up a bit.  I made a 2-disk mirror for OS, and a
> 4-disk RAID-10 for the NBU catalog.  This did not work well.  It turns
> out
> the internal LSI RAID controller is brain dead (as are most low-end raid
>
> controllers) and was doing a round robin read on what actually appeared
> to be a RAID-0+1.  So, my performance was actually worse than it was on
> a
> single spindle.
>
> As a temporary measure, I moved the catalog to SAN space.  It screamed.
> SAN was the best performance I have seen out of the catalog (see two
> HBAs
> above), but I wasn't too comfortable putting my catalog on SATA disk,
> SAN or not, so this was just a temporary home while I converted the
> 4-disk "RAID-10" to RAID-5.  Once this was complete, I had acceptible
> speeds again (above the minimum 10MB/s for my HP LTO2 drives).
>
>

-- 
Joe Royer / SysAdmin / Digital Motorworks / 512-692-1028


This message and any attachments are intended only for the use of the addressee 
and may contain information that is privileged and confidential. If the reader 
of the message is not the intended recipient or an authorized representative of 
the intended recipient, you are hereby notified that any dissemination of this 
communication is strictly prohibited. If you have received this communication 
in error, please notify us immediately by e-mail and delete the message and any 
attachments from your system.
_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu