Re: Benchmarking

From: Tatsuo Ishii <t-ishii(at)sra(dot)co(dot)jp>
To: romero(at)kde(dot)org
Cc: pgsql-hackers(at)postgresql(dot)org
Subject: Re: Benchmarking
Date: 2001-06-27 01:11:23
Message-ID: 20010627101123S.t-ishii@sra.co.jp
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

> I need to check the scalability of a machine with postgresql and Im doing it
> with pgbench but Im getting values with a variation of a 40% with the same
> pgbench call...

You might be looking at the effect of the kernel buffer cache. Try run
pgbench several times with same settings. Another point is how many
transactions pgbench runs (-t option). More transactions would give
more statble results. Here is my small script to run pgbench. I
usually run it 2 or 3 times and take only the last run result.

#! /bin/sh
pgbench -i -s 2 test
for i in 1 2 4 8 16 32 64 128
do
t=`expr 640 / $i`
pgbench -t $t -c $i test
echo "===== sync ======"
sync;sync;sync;sleep 10
echo "===== sync done ======"
done

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Thomas T. Veldhouse 2001-06-27 01:27:03 Bug in createlang?
Previous Message Reinoud van Leeuwen 2001-06-26 23:25:23 Re: functions returning records