Need Some Suggestions

From: "Lane Van Ingen" <lvaningen(at)esncc(dot)com>
To: <pgsql-performance(at)postgresql(dot)org>
Subject: Need Some Suggestions
Date: 2005-10-06 21:29:42
Message-ID: EKEMKEFLOMKDDLIALABIMEHACDAA.lvaningen@esncc.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

I have an application that is prone to sudden, unscheduled high bursts of
activity, and
I am finding that the application design permits me to detect the activity
bursts within
an existing function. The bursts only affect 3 tables, but degradation
becomes apparent
after 2,000 updates, and significant after 8,000 updates.

I already know that a plain vacuum (without full, analyze, or free options)
solves my
problem. Since vacuum is classified in the documentation as an SQL command,
I tried to
call it using a trigger function on one the tables (they all have roughly
the same insert
/ update rate). However, I just found out that vacuum cannot be called by a
function.
Vacuums done by a scheduler at 3AM in the morning are adequate to handle my
non-peak
needs otherwise.

autovacuum sounds like it would do the trick, but I am on a WINDOWS 2003
environment, but
I have Googled up messages that it still has various problems (in Windows)
which won't be
resolved until 8.1 is out. But I have a problem NOW, and the application is
deployed
around the world.

QUESTION:
Is there anyway anyone knows of to permit me to execute an operating
system program
(even vacuumdb) or possibly to add a C function to the library which would
allow me to
do this (I am not a C programmer, but have access to some persons who are)?

Very important to me for performance reasons.

Does anybody have some suggestions on the best path for me to take?

Responses

Browse pgsql-performance by date

  From Date Subject
Next Message Martijn van Oosterhout 2005-10-06 21:40:17 Re: [PERFORM] A Better External Sort?
Previous Message Tom Lane 2005-10-06 20:25:11 Re: [HACKERS] A Better External Sort?