pg.ddx.io pgsql-performance@postgresql.org mailing list archive
help / color / mirror / Atom feedAcceptable level of over-estimation?
2+ messages / 2 participants
[nested] [flat]
* Acceptable level of over-estimation?
@ 2007-09-24 19:37 Carlo Stonebanks <stonec.register@sympatico.ca>
0 siblings, 1 reply; 2+ messages in thread
From: Carlo Stonebanks @ 2007-09-24 19:37 UTC (permalink / raw)
To: pgsql-performance
Is there a rule of thumb about when the planner's row estimates are too
high? In particular, when should I be concerned that planner's estimated
number of rows estimated for a nested loop is off? By a factor of 10? 100?
1000?
Carlo
^ permalink raw reply [nested|flat] 2+ messages in thread
* Re: Acceptable level of over-estimation?
@ 2007-09-24 20:18 Gregory Stark <stark@enterprisedb.com>
parent: Carlo Stonebanks <stonec.register@sympatico.ca>
0 siblings, 0 replies; 2+ messages in thread
From: Gregory Stark @ 2007-09-24 20:18 UTC (permalink / raw)
To: Carlo Stonebanks <stonec.register@sympatico.ca>; +Cc: pgsql-performance
"Carlo Stonebanks" <stonec.register@sympatico.ca> writes:
> Is there a rule of thumb about when the planner's row estimates are too high?
> In particular, when should I be concerned that planner's estimated number of
> rows estimated for a nested loop is off? By a factor of 10? 100? 1000?
Not really. It's a big enough difference for the planner to make a bad
decision or it isn't. But if you pressed me I would say a factor of 10 is bad.
A factor of 2 is inevitable in some cases.
--
Gregory Stark
EnterpriseDB http://www.enterprisedb.com
^ permalink raw reply [nested|flat] 2+ messages in thread
end of thread, other threads:[~2007-09-24 20:18 UTC | newest]
Thread overview: 2+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2007-09-24 19:37 Acceptable level of over-estimation? Carlo Stonebanks <stonec.register@sympatico.ca>
2007-09-24 20:18 ` Gregory Stark <stark@enterprisedb.com>
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox