| From: | Christopher Browne <cbbrowne(at)libertyrms(dot)info> |
|---|---|
| To: | pgsql-novice(at)postgresql(dot)org |
| Subject: | Re: Number of rows in a table |
| Date: | 2003-09-03 03:15:11 |
| Message-ID: | 60isoa4l6o.fsf@dev6.int.libertyrms.info |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-novice |
ron(dot)l(dot)johnson(at)cox(dot)net (Ron Johnson) writes:
> Well yes, but in most other databases, there is a system table that
> stores the approximate number of records in each table, and querying
> that system table is a *lot* faster than sequentially reading a 100M
> row table.
That relation would be pg_class, the domain is called "reltuples."
It is only about as accurate as the last vacuum and/or analyze has
made it, so if you don't fairly regularly vacuum tables, the
approximation may not be very good. (Which is an argument in favor of
vacuuming fairly often...)
--
select 'cbbrowne' || '@' || 'libertyrms.info';
<http://dev6.int.libertyrms.com/>
Christopher Browne
(416) 646 3304 x124 (land)
| From | Date | Subject | |
|---|---|---|---|
| Next Message | glenn | 2003-09-03 05:07:35 | setting next value of an established sequence |
| Previous Message | Tyler Colbert | 2003-09-03 01:44:51 | Re: using dates in pgsql |