From: | Tom Harkaway <tomh(at)hvpa(dot)com> |
---|---|
To: | "pgsql-performance(at)postgresql(dot)org" <pgsql-performance(at)postgresql(dot)org> |
Subject: | 8.4 to 9.2 migration performance |
Date: | 2013-07-08 16:21:31 |
Message-ID: | 5455ED1E62437A4696143D543F150567DA2C26@S04-MBX01-07.s04.local |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-performance |
I am migrating a Postgres 8.4 installation on a dedicated server to Postgres 9.2 running on a Virtual Machine. A sample query that run in 10 minutes on the 8.4 installation take 40 minutes on the 9.2 installation.
Current Server, Postgres 8.4
* 6-core, 3GHz AMD system
* 12GB of RAM
* 4 SATA drive RAID-1 storage
* Mandriva OS
* SQL encoding and 'C' collation
Virtual Machine, Postgres 9.2 ( two different systems)
* 4-core, 3Ghz Intel system
* 12GB or RAM
* SAS storage on one, and 4-SATA drive RAID-10 system on second
* CentOS 6.3 OS
* UTF-8 encoding, and I have tried both 'C' and en_US collation
The first VM is at a local Data Center and the second in on a dedicated server in my office. Both give similar results.
The data, indexes and constraints have all been successfully migrated to the new system.
I have tuned the VM systems using pgtune with no significant before and after difference.
The 'explain' output for the query is very different between the two systems.
It seems like I am missing some simple step for there to be such a huge performance difference.
Any suggestions on what else to text/check would be very much appreciated.
Tom
From | Date | Subject | |
---|---|---|---|
Next Message | Jeison Bedoya | 2013-07-08 16:22:12 | Process in state BIND, authentication, PARSE |
Previous Message | Jeison Bedoya | 2013-07-08 16:14:52 | Performance autovaccum |