Re: random_page_costs - are defaults of 4.0 realistic for SCSI RAID 1

From: Mark Mielke <mark(at)mark(dot)mielke(dot)cc>
To: "Decibel!" <decibel(at)decibel(dot)org>
Cc: Scott Marlowe <scott(dot)marlowe(at)gmail(dot)com>, Gregory Stark <stark(at)enterprisedb(dot)com>, Luke Lonergan <llonergan(at)greenplum(dot)com>, Carlo Stonebanks <stonec(dot)register(at)sympatico(dot)ca>, pgsql-performance(at)postgresql(dot)org
Subject: Re: random_page_costs - are defaults of 4.0 realistic for SCSI RAID 1
Date: 2007-09-11 21:50:24
Message-ID: 46E70DA0.6070008@mark.mielke.cc
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

Decibel! wrote:
> On Mon, Sep 10, 2007 at 06:22:06PM -0400, Mark Mielke wrote:
>
>> In my case, I set effective_cache_size to 25% of the RAM available to
>> the system (256 Mbytes), for a database that was about 100 Mbytes or
>> less. I found performance to increase when reducing random_page_cost
>> from 4.0 to 3.0.
>>
> Just for the record, effective_cache_size of 25% is *way* too low in
> most cases, though if you only have 1GB setting it to 500MB probably
> isn't too far off.
>
> Generally, I'll set this to however much memory is in the server, minus
> 1G for the OS, unless there's less than 4G of total memory in which case
> I subtract less.
>
Agree. My point was only that there are conflicting database
requirements, and that one setting may not be valid for both. The
default should be whatever is the most useful for the most number of
people. People who fall into one of the two extremes should know enough
to set the value based on actual performance measurements.

Cheers,
mark

--
Mark Mielke <mark(at)mielke(dot)cc>

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Decibel! 2007-09-11 22:52:42 Re: [Again] Postgres performance problem
Previous Message Decibel! 2007-09-11 21:31:05 Re: random_page_costs - are defaults of 4.0 realistic for SCSI RAID 1