From: | Jean-Christophe ARNU (JX) <jc(dot)arnu(at)wanadoo(dot)fr> |
---|---|
To: | pgsql-admin(at)postgresql(dot)org |
Subject: | Timestamps and performance problems on queries. |
Date: | 2002-04-10 08:12:58 |
Message-ID: | 20020410041258.345a35ec.jc.arnu@wanadoo.fr |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-admin |
Hello all.
I've a performance problem on specific requests :
When I use timestamps + interval in where clauses, query performance is slowed
down by a factor of 20 or 30!!!! For exemple :
select timestamp,value
from measure
where timestamp<now() and timestamp>(now() - '1 hour'::interval)
is 20 to 30 times longer than
select timestamp,value
from measure
where timestamp<'2002-04-10 10:00' and timestamp>='2002-04-10 9:00';
So where is the bottleneck?
A paradigm seems that now() and (now() - '1hour'::interval) is evaluated for
each row comparison... Am I right? Thus is there a way to make SQL interpreter
evaluate this by rewriting them before launching any comparisons?
Or do I have to rewrite all my application queries and calculate each time
now() and interval predicates?
Thanks by advance
--
Jean-Christophe ARNU
s/w developer
Paratronic France
> (Pas Adabase que j'ai en version démo sur la distribution Suze).
Et après ça, allez expliquer que les linuxiens sont pas alcolos... ;-)
à quand une distribution Kro v1.6.64 ou Pastis v5.1?
-+- Fantec In Guide du linuxien pervers - "C'est l'heure de l'apéro ?" -+-
From | Date | Subject | |
---|---|---|---|
Next Message | Jean-Christophe ARNU (JX) | 2002-04-10 08:22:15 | Timestamps and performances problems |
Previous Message | Stephan Szabo | 2002-04-09 15:53:25 | Re: table dump |