Bacula-users

Re: [Bacula-users] Throughput speed to SAS LTO4

2010-07-15 23:01:56
Subject: Re: [Bacula-users] Throughput speed to SAS LTO4
From: Rory Campbell-Lange <rory AT campbell-lange DOT net>
To: richard <richard AT sauce.co DOT nz>
Date: Fri, 16 Jul 2010 03:59:19 +0100
Hi Richard

On 16/07/10, richard (richard AT sauce.co DOT nz) wrote:
> Rory Campbell-Lange wrote:
> 
> > Currently I'm getting a throughput of around 10MB/sec to tape, which
> > seems pretty low. iotop is showing reading off disk at around
> > 17-20MB/sec and one of the 8 cores on this Xeon E5520 (2267MHz) is
> > running at 100%. Hardware compression is disabled on the tape changer.
> >
> > Are there any obvious ways to improve this? Does the problem look like
> > gzip, and if so is pigz (parallel gzip) an option?
> 
> Bacula compression is single threaded, so could we be the problem.
> 
> Why are you not using the hardware compression? It adds no throughput 
> overhead.

I've enabled hardware compression and throughput speed has improved to
around 80MB/sec which is better but still not fantastic. It is coming
off disk at around 90-100MB/sec so the diffence must be partly down to
the hw compression.

As a matter of interest, is hardware compression an issue if the data
needs to come off these tapes onto another LTO4 drive? It is important
that this data can be read for a long time into the future.

Kind regards
Rory

--
Rory Campbell-Lange
rory AT campbell-lange DOT net

------------------------------------------------------------------------------
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users

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