On Mon, 2 Jul 2007, Tom Lane wrote:
>> # wal_buffers = 1MB
> Is there really evidence in favor of such a high setting for this,
> either?
I noticed consistant improvements in throughput on pgbench results with
lots of clients going from the default to 256KB, flatlining above that; it
seemed sufficiently large for any system I've used. I've taken to using
1MB anyway nowadays because others suggested that number, and it seemed to
be well beyond the useful range and thus never likely to throttle
anything. Is there any downside to it being larger than necessary beyond
what seems like a trivial amount of additional RAM?
>> # checkpoint_segments = 8 to 16 if you have the disk space (0.3 to 0.6 GB)
> This seems definitely too small --- for write-intensive databases I like
> to set it to 30 or so, which should eat about a GB if I did the
> arithmetic right.
You did--I approximate larger values in my head by saying 1GB at 30
segments and scaling up from there. But don't forget this is impacted by
the LDC change, with the segments expected to be active now
(2 + checkpoint_completion_target) * checkpoint_segments + 1
so with a default install setting the segments to 30 will creep that up to
closer to a 1.2GB footprint.
--
* Greg Smith gsmith@gregsmith.com http://www.gregsmith.com Baltimore, MD