On Thu, 2009-05-07 at 20:36 +0200, Dimitri wrote:
> I've simply restarted a full test with hashjoin OFF. Until 32
> concurrent users things are going well. Then since 32 users response
> time is jumping to 20ms, with 64 users it's higher again, and with 256
> users reaching 700ms, so TPS is dropping from 5.000 to ~200..
>
> With hashjoin ON it's not happening, and I'm reaching at least 11.000
> TPS on fully busy 32 cores.
Much better to stick to the defaults.
Sounds like a problem worth investigating further, but not pro bono.
> About scalability issue - there is one on 8.3.7, because on 32 cores
> with such kind of load it's using only 50% CPU and not outpassing
> 6.000 TPS, while 8.4 uses 90% CPU and reaching 11.000 TPS..
Yeh, small changes make a big difference. Thanks for the info.
How does MySQL perform?
--
Simon Riggs www.2ndQuadrant.com
PostgreSQL Training, Services and Support