Re: Different execution time for same plan - Mailing list pgsql-hackers

From Kevin Grittner
Subject Re: Different execution time for same plan
Date
Msg-id 4DECB091020000250003E1D8@gw.wicourts.gov
Whole thread Raw
In response to Different execution time for same plan  (Nick Raj <nickrajjain@gmail.com>)
List pgsql-hackers
First off, this is posted to the wrong list -- this list is for
discussion of development of the PostgreSQL product.  There is a
list for performance questions where this belongs:
pgsql-performance@postgresql.org.  I'm moving this to the
performance list with a blind copy to the -hackers list so people
know where the discussion went.

Nick Raj <nickrajjain@gmail.com> wrote:

> When i execute the query first time, query takes a quite longer
> time but second time execution of the same query takes very less
> time (despite execution plan is same)

> Why the same plan giving different execution time? (Reason may be
> data gets buffered (cached) for the second time execution) Why
> there is so much difference?

Because an access to a RAM buffer is much, much faster than a disk
access.

> Which option will be true?

It depends entirely on how much of the data needed for the query is
cached.  Sometimes people will run a set of queries to "warm" the
cache before letting users in.

> MY postgresql.conf file having setting like this (this is original
> setting, i haven't modify anything)

> shared_buffers = 28MB

> #work_mem = 1MB                # min 64kB
> #maintenance_work_mem = 16MB        # min 1MB

If you're concerned about performance, these settings (and several
others) should probably be adjusted:

http://wiki.postgresql.org/wiki/Tuning_Your_PostgreSQL_Server

-Kevin


pgsql-hackers by date:

Previous
From: Heikki Linnakangas
Date:
Subject: Re: BLOB support
Next
From: Robert Haas
Date:
Subject: Re: reducing the overhead of frequent table locks - now, with WIP patch