[prev in list] [next in list] [prev in thread] [next in thread] 

List:       bacula-devel
Subject:    Re: [Bacula-devel] [Bacula-users] Bacula performance with a 64512
From:       Arno Lehmann <al () its-lehmann ! de>
Date:       2007-09-24 20:13:09
Message-ID: 46F81A55.6030504 () its-lehmann ! de
[Download RAW message or body]

Hi,

20.09.2007 14:53,, Kern Sibbald wrote::
...
>> The thing is if bacula is too slow in writing to tapes this might be a
>> showstopper for large environments because this a) wears the drive
>> ("shoe-shining") and b) will make backups run way too long.
> 
> I think Ralf has shown that Bacula is not too slow writing tapes with the 
> default size.  I don't say that it couldn't be faster, but when it comes to 
> backups, I am not a cowboy.  I believe in being conservative and doing *very* 
> careful testing.  

I've got something to report, too: With the AIT-5 system I was testing 
with, Bacula, with its default block size, got almost 20MB/s. This is 
a) much better than what dd managed with this block size, and b) close 
enough to the published specs of 24MB/s to prevent shoe-shine.

Especially a) makes me wonder - I can't recall seeing this sort of 
difference yet.

Still, I'll try to see what we can achieve with compression and 
compressible data. I don't believe any tape drive manufacturer their 
published compression ratios, but 1.5:1 is not uncomman, and that 
would require 30-40 MB/s transfer rate to the tape drive. I'm curious 
if that's possible with the default block size :-)

Arno

-- 
Arno Lehmann
IT-Service Lehmann
www.its-lehmann.de

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Bacula-devel mailing list
Bacula-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-devel
[prev in list] [next in list] [prev in thread] [next in thread]