Re: Select count(*), the sequel

16.10.10 19:51, Mladen Gogala написав(ла):
> There was some doubt as for the speed of doing the select count(*) in
> PostgreSQL and Oracle.
> To that end, I copied the most part of the Oracle table I used before
> to Postgres. Although the copy
> wasn't complete, the resulting table is already significantly larger
> than the table it was copied from. The result still shows that Oracle
> is significantly faster:

Hello.

Did you vacuum postgresql DB before the count(*). I ask this because
(unless table was created & loaded in same transaction) on the first
scan, postgresql has to write hint bits to the whole table. Second scan
may be way faster.

Best regards, Vitalii Tymchyshyn