Liam Slusser wrote:
> I've been trying to replicate a database but each time I replication it the
> performance of the copy is about 100 times slower (~100ms to ~8 seconds for
> the same query). The only way I have found to replicate it and keep the
> same performance is doing a hotcopy of the database.
>
> Please note I didn't design this database, I just have to support it.
>
> $ uname -a
> Linux hostname 2.6.20-gentoo-r8-5 #2 SMP Wed Aug 1 19:43:33 CDT 2007 x86_64
> Intel(R) Xeon(R) CPU 5130 @ 2.00GHz GenuineIntel GNU/Linux
>
> $ psql --version
> psql (PostgreSQL) 8.1.5
>
> So the original database, cmpub, works great. But when I do a pg_dump and
> import it to a test database on the same server, or another server for that
> matter, the performance is awful.
>
> Here is how I did the test....
>
> Create test database:
>
> $ ./createdb --template template1 --encoding UNICODE liam
> $ ./pg_dump cmpub | ./psql liam
>
> Run a vacuum full:
> liam=# vacuum full verbose;
You need to analyze, not vacuum full. pg_dump doesn't include any
analyze statements, you need to do that manually.
--
Postgresql & php tutorials
http://www.designmagick.com/