On Fri, 14 Nov 2003 20:38:33 +1100 (EST)
Slavisa Garic <Slavisa(dot)Garic(at)infotech(dot)monash(dot)edu(dot)au> wrote:
> Any help would be greatly appreciated even pointing me in the right
> direction where to ask this question. By the way I designed the
> database this way as my application that uses PGSQL a lot during the
> execution so there was a huge need for fast SELECTs. Our experiments
> are getting larger and larger every day so fast inserts would be good
> as well.
First, you need to upgrade to 7.3.4, 7.4 is prefable if a dump/reload is
not too bad.
Standard set of questions:
1. Any foreign keys
2. are these inserts batched into transactions
3. CPU usage?
5. PG config? [shared_buffers, effective_cache_size, etc]
6. IO saturation?
Also, try searching the archives. lots of juicy info there too.
Jeff Trout <jeff(at)jefftrout(dot)com>
In response to
pgsql-performance by date
|Next:||From: Christopher Browne||Date: 2003-11-14 14:13:48|
|Subject: Re: Help with count(*)|
|Previous:||From: Nick Barr||Date: 2003-11-14 10:11:44|
|Subject: IDE Hardware RAID Controller|