Re: shared_buffers/effective_cache_size on 96GB server

On Wed, Oct 10, 2012 at 12:12 AM, Strahinja Kustudić
<strahinjak(at)nordeus(dot)com> wrote:
> Hi everyone,
>
> I have a Postgresql 9.1 dedicated server with 16 cores, 96GB RAM and RAID10
> 15K SCSI drives which is runing Centos 6.2 x64.

How many drives in the RAID?

> This server is mainly used
> for inserting/updating large amounts of data via copy/insert/update
> commands, and seldom for running select queries.

Are there a lot of indexes?

>
> Here are the relevant configuration parameters I changed:
>
> shared_buffers = 10GB
> effective_cache_size = 90GB
> work_mem = 32MB
> maintenance_work_mem = 512MB
> checkpoint_segments = 64
> checkpoint_completion_target = 0.8
>
> My biggest concern are shared_buffers and effective_cache_size, should I
> increase shared_buffers and decrease effective_cache_size?

Are you experiencing performance problems? If so, what are they?

> I read that
> values above 10GB for shared_buffers give lower performance, than smaller
> amounts?

There are reports that large shared_buffers can lead to latency
spikes. I don't know how sensitive your work load is to latency,
though. Nor how much those reports apply to 9.1.

>
> free is currently reporting (during the loading of data):
>
> $ free -m
> total used free shared buffers cached
> Mem: 96730 96418 311 0 71 93120
> -/+ buffers/cache: 3227 93502
> Swap: 21000 51 20949
>
> So it did a little swapping, but only minor,

The kernel has, over the entire time the server has been up, found 51
MB of process memory to swap. That doesn't really mean anything. Do
you see active swapping going on, like with vmstat?

Cheers,

Jeff