Veritas-bu

[Veritas-bu] LTO-3 throughput expectations

2006-12-21 13:21:17
Subject: [Veritas-bu] LTO-3 throughput expectations
From: BBahnmiller at pier1.com (Bahnmiller, Bryan)
Date: Thu, 21 Dec 2006 12:21:17 -0600
  What's your fragment size on the storage unit? If you have a 1 GB
fragment size, the tape drive basically spins up to speed and then slows
down again, never having really gotten up to full speed. When I wrote 1
GB fragments, the tape drive would write for 6 seconds before slowing
down, stopping, and then repositioning for the next file marker.

  I found that in our setup, 10 GB fragments actually worked best for
LTO3.

Bryan  

> -----Original Message-----
> From: veritas-bu-bounces at mailman.eng.auburn.edu 
> [mailto:veritas-bu-bounces at mailman.eng.auburn.edu] On Behalf 
> Of Steve Kirkpatrick
> Sent: Thursday, December 21, 2006 11:56 AM
> To: Austin Murphy
> Cc: veritas-bu at mailman.eng.auburn.edu
> Subject: Re: [Veritas-bu] LTO-3 throughput expectations
> 
> Thanks for the info Austin.  You pretty much confirm what I 
> have been reading.
> 
> Anyone else have an idea of why I can only seem to drive the 
> tape drives at 12.5MB/sec when duplicating tapes on the 
> master server?  The server itself is not loaded during this 
> activity.  It seems that bpduplicate should be able to read 
> at max speed and write at max speed simultaneously especially 
> since the drives and server are connected over a 2gb/sec SAN.
> 
> I guess the long-term solution is to go with a disk-based 
> device (such as the Data Domain) for the initial backups and 
> then archive to tape from there.
> 
> Also, what about the bpduplicate performance when duplicating 
> multiplexed tapes?  How about restores from the same?  Am I 
> just moving the bottleneck from one place to another?
> 
> Any links to info that will educate me on the subject are appreciated.
> 
> Thanks again,
> Steve.
> 
> > -----Original Message-----
> > From: Austin Murphy [mailto:austin.murphy at gmail.com]
> > Sent: Thursday, December 21, 2006 7:24 AM
> > To: Steve Kirkpatrick
> > Cc: veritas-bu at mailman.eng.auburn.edu
> > Subject: Re: [Veritas-bu] LTO-3 throughput expectations
> > 
> > 
> > On 12/21/06, Steve Kirkpatrick <skirkpatrick at ghx.com> wrote:
> > > NB5.0MP6
> > > Master server: Sun V440 w/ 4x1.2GHZ SPARC III, 8GB RAM
> > running Solaris 10
> > > Sun L500 library with four HPUltrium LTO-3 drives connected
> > to master server
> > > via 2GB SAN.
> > >
> > > I can't seem to get more than 12.5MB/sec throughput on
> > these LTO-3 drives.  That
> > > is about the same as I am getting on another server's
> > SCSI-connected LTO-1 drives.
> > > I see the same performance whether I am duplicating tapes
> > using bpduplicate or
> > > when I am backing up clients over a gigabit network.
> > >
> > > After reading all of the recent threads concerning tape
> > drive performance,
> > > shoeshining, etc, I can't believe that the master server is
> > only able to feed
> > > 12.5MB/sec of data to these drives.
> > 
> > You bottleneck is most likely the network and/or the source 
> system.  A 
> > gigabit network maxes out at 35 MB/sec in some cases. 
> Getting better 
> > than 50 MB/sec is rare.
> > 
> > > Some of the previous threads on the subject suggested
> > multiplexing jobs across
> > > drives but I worry that doing so will cause my duplicate
> > jobs to take longer as well
> > > as hamper restore performance.  Need I be worried about that?
> > 
> > If you don't multiplex, then your max speed for the tape 
> will be the 
> > max speed of the server yo are backing up.  If you factor in the 
> > compression ratio, you will have an even more difficult time 
> > maintaining the minimum sptreaming speed.
> > 
> > You almost certainly need to use multiplexing and cut down 
> the number
> > of active drives in your storage unit.   Try 1 drive and MPX=4.   If
> > you can add a network link you should be able to add a 
> second drive to 
> > your storage unit.
> > 
> > Austin
> > 
> 
> _______________________________________________
> Veritas-bu maillist  -  Veritas-bu at mailman.eng.auburn.edu 
> http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
> 


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