Re: Timestamps and performances problems
От
Jean-Christophe ARNU (JX)
Тема
Re: Timestamps and performances problems
Дата
Msg-id
20020410104425.59ce6791.jc.arnu@free.fr
Список
Дерево обсуждения
Timestamps and performances problems Jean-Christophe ARNU (JX) <jc.arnu@free.fr>
Re: Timestamps and performances problems Tom Lane <tgl@sss.pgh.pa.us>
Please help Tariq Muhammad <tmuhamma@mail.libertyrms.com>
Le Wed, 10 Apr 2002 09:27:09 -0400 "Dan Langille" me disait que : > On 10 Apr 2002 at 9:13, JX wrote: > > > Le Wed, 10 Apr 2002 09:06:55 -0400 > > "Dan Langille" me disait que : > > > > > On 10 Apr 2002 at 11:51, Gaetano Mendola wrote: > > > > > > > "Jean-Christophe ARNU (JX)" wrote: > > > > > 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() - '1 hour'::interval) > > > > > > Try where timestamp(now() - '1 > > > hour'::interval)::timestemp. > > > > What's the difference with the syntax above? It takes he same time > > than the query above. Bounded timestamps with "real" ISO timestamps > > strings are always up to about 200 times faster (with extensive test > > proof). > > It casts the value to a timestamp. I would prefer to discuss this on- > list. Okaye, but what's the incidence on preformance issues? Casting should only insure that given string is to be taken as a timestamp isn't it? Does it make an "instanciation" of the timestamp to be that would be applied for comparision clauses? Thanks -- Jean-Christophe ARNU s/w developer Paratronic France «Dès que je clique sur "mount", il ne fait rien et le cdrom reste "unmount". Quel est le pb ?» -+- Popol in Guide du linuxien pervers : "De l'avantage des interfaces..." -+-
В списке pgsql-admin по дате отправления