[DRBD-user] Performance regression with DRBD 8.3.12 and newer

Lars Ellenberg lars.ellenberg at linbit.com
Thu Jun 14 10:52:55 CEST 2012

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


On Wed, Jun 13, 2012 at 08:31:05PM +0200, Dirk wrote:
> Sorry, but I am not deeply into kernel details, so it is not easy to
> follow the thread and filter a result.
> >The performance degradation you experience is simply that your
> >config did not disable these, and if compiled against the right
> >headers, they are used and effective.
> What exactly do I have to do to avoid the perf degradation? Do I
> have to compile a modified kernel and compile drbd.ko against that?
> If yes, what changes to the kernel config?

 * there is no performance degradation in DRBD
 * what you perceive as performance degradation in DRBD is
   the difference in performance between
   barriers and flushes being used vs. being not supported
 * to "get back" your performance,
   explicitly disable barriers and flushes in drbd.


Cheers,

-- 
: Lars Ellenberg
: LINBIT | Your Way to High Availability
: DRBD/HA support and consulting http://www.linbit.com



More information about the drbd-user mailing list