Re: Query using SeqScan instead of IndexScan - Mailing list pgsql-performance

From Jim Nasby
Subject Re: Query using SeqScan instead of IndexScan
Date
Msg-id 393CBC49-14DB-4CBE-B824-7541CBD7BACC@pervasive.com
Whole thread Raw
In response to Re: Query using SeqScan instead of IndexScan  (Josh Berkus <josh@agliodbs.com>)
List pgsql-performance
On Apr 2, 2006, at 6:30 PM, Josh Berkus wrote:
>> But just as a follow up question to your #1 suggestion, I have 8 GB
>> of ram in my production server. You're saying to set the
>> effective_cache_size then to 5 GB roughly? Somewhere around 655360?
>> Currently it is set to 65535. Is that something that's OS dependent?
>> I'm not sure how much memory my server sets aside for disk caching.
>
> Yes, about.  It's really a judgement call; you're looking for the
> approximate
> combined RAM available for disk caching and shared mem.  However,
> this is
> just used as a way of estimating the probability that the data you
> want is
> cached in memory, so you're just trying to be order-of-magnitude
> accurate,
> not to-the-MB accurate.

FWIW, I typically set effective_cache_size to the amount of memory in
the machine minus 1G for the OS and various other daemons, etc. But
as Josh said, as long as your somewhere in the ballpark it's probably
good enough.
--
Jim C. Nasby, Sr. Engineering Consultant      jnasby@pervasive.com
Pervasive Software      http://pervasive.com    work: 512-231-6117
vcard: http://jim.nasby.net/pervasive.vcf       cell: 512-569-9461



pgsql-performance by date:

Previous
From: "Marc Morin"
Date:
Subject: Re: Query runs too long for indexed tables
Next
From: Jim Nasby
Date:
Subject: Re: Query using SeqScan instead of IndexScan