From: | Craig Ringer <ringerc(at)ringerc(dot)id(dot)au> |
---|---|
To: | Misa Simic <misa(dot)simic(at)gmail(dot)com> |
Cc: | Andrew Dunstan <andrew(at)dunslane(dot)net>, "pgsql-performance(at)postgresql(dot)org" <pgsql-performance(at)postgresql(dot)org> |
Subject: | Re: Paged Query |
Date: | 2012-07-09 23:50:29 |
Message-ID: | 4FFB6E45.50708@ringerc.id.au |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-performance |
On 07/10/2012 06:24 AM, Misa Simic wrote:
> Hi Andrew,
>
> Sure... We are sending data in Json to clients
>
> {
> total:6784,
> data:[50 rows for desired page]
> }
>
> SELECT count(*) FROM table - without where, without joins used to have
> bad performance... However, in real scenario we have never had the case
> without joins and where... Join columns are always indexed, and we
> always try to put indexes on columns what will mostly be used in where
> in usual queries...
When/if you do need a count of a single table without any filters, a
common trick is to use table statistics to return an approximation. If
your autovaccum is running regularly it's usually a very good
approximation, too.
Sounds like this hack may become unnecessary in 9.2 though.
--
Craig Ringer
From | Date | Subject | |
---|---|---|---|
Next Message | Jeff Janes | 2012-07-10 00:34:14 | Re: Paged Query |
Previous Message | Craig Ringer | 2012-07-09 23:48:28 | Re: Paged Query |