[stgt] Performance Issue

ronnie sahlberg ronniesahlberg at gmail.com
Thu Feb 9 21:27:27 CET 2012


We removed the fsycn() a short while ago from WRITE6
and only do a single fsync() once the backup is complete and the
initator does the final WRITEFILEMARK.

With this change you should not have a huge impact any more from the
fsync() calls.

Could you do a test and compare with the latest version of STGT that
has this change ?


regards
ronnie sahlberg

On Fri, Feb 10, 2012 at 5:37 AM, Mark Lehrer <mark at knm.org> wrote:
>> You probably shouldnt remove fsync() completely,
>> but make sure that you have fsync() called at least from the
>> WriteFilemark emulation.
>
>
> In my test lab here I added a hack to turn off both O_DIRECT and fsync.
> Would anyone be interested in a per-target config option for this; maybe
> something like UnsafeWriteCaching that makes it clear that it is not a
> production option?
>
> This made like a 1000x improvement in disk performance for a set of Oracle
> VM's that we use for temporary copies of our full customer data.
>
> Thanks,
> Mark
>
>
--
To unsubscribe from this list: send the line "unsubscribe stgt" in
the body of a message to majordomo at vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html



More information about the stgt mailing list