From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Magnus Hagander <magnus(at)hagander(dot)net> |
Cc: | Wei Yan <weiyan1(at)gmail(dot)com>, pgsql-performance(at)postgresql(dot)org |
Subject: | Re: Help: how to speed up query after db server reboot |
Date: | 2009-09-03 14:32:04 |
Message-ID: | 14128.1251988324@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-performance |
Magnus Hagander <magnus(at)hagander(dot)net> writes:
> On Wed, Sep 2, 2009 at 00:01, Wei Yan<weiyan1(at)gmail(dot)com> wrote:
>> Looks like after postgres db server reboot, first query is very slow
>> (10+mins). After the system cache built, query is pretty fast.
>> Now the question is how to speed up the first query slow issue?
> Schedule a run of a couple of representative queries right as the
> database has started? That should pre-populate the cache before your
> users get there, hopefully.
I wonder if VACUUMing his key tables would be a good answer. I bet that
a lot of the problem is swapping in indexes in a slow random-access
fashion. In recent-model Postgres, VACUUM will do a sequential scan of
the indexes (at least for btree) which should be a much more efficient
way of bringing them into kernel cache.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Merlin Moncure | 2009-09-03 14:49:36 | Re: partition query using Seq Scan even when index is present |
Previous Message | Tguru | 2009-09-03 13:52:53 | Re: SAAS and MySQL |