> Thanks John.
>
> Well as I mentioned. I have a Dual AMD Opteron 64 2.4ghz, 15k rpm
SCSI
> Disks, 4GB of memory.
> Disks are pretty fast and memory should be more than enough. Currently
> we dont have many concurrent connections.
>
> I run PG 8.0.1 on Fedora Core 3
>
> When I now run the batch job, one CPU runs in the 80-90% the other in
> 5-10% max.
If possible, split up your job into two workloads that can be run
concurrently. Open up two connections on the client, one for each
workload. To be 100% sure they get delegated to separate processors,
let the first connection start working before opening the second one
(should be easy enough to test from the terminal)...this should pretty
much guarantee your batch processes get delegated to different
processors.
The beauty of pg is that it lets the o/s handle things that should be
the o/s's job...don't forget the bgwriter/stats collector processes are
also in the mix. Your o/s is probably already doing a pretty good job
delegating work already.
Even with fast disks, batch data management jobs are rarely cpu bound,
so you might not see much difference in the total run time (spitting
your batch might actually increase the run time, or reduce
responsiveness to other connections). Never hurts to test that though.
Merlin
