The response time is not progressive, it's simply jumping, it's likely
since 16 sessions there is a sort of serialization happening
somewhere.. As well on 16 sessions the throughput in TPS is near the
same as on 8 (response time is only twice bigger for the moment), but
on 32 it's dramatically dropping down..
Rgds,
-Dimitri
On 5/19/09, Simon Riggs <simon@2ndquadrant.com> wrote:
>
> On Tue, 2009-05-19 at 14:00 +0200, Dimitri wrote:
>
>> I may confirm the issue with hash join - it's repeating both with
>> prepared and not prepared statements - it's curious because initially
>> the response time is lowering near ~1ms (the lowest seen until now)
>> and then once workload growing to 16 sessions it's jumping to 2.5ms,
>> then with 32 sessions it's 18ms, etc..
>
> Is it just bad all the time, or does it get worse over time?
>
> Do you get the same behaviour as 32 sessions if you run 16 sessions for
> twice as long?
>
> --
> Simon Riggs www.2ndQuadrant.com
> PostgreSQL Training, Services and Support
>
>