Re: Huge iowait during checkpoint finish

From: Scott Marlowe <scott(dot)marlowe(at)gmail(dot)com>
To: Greg Smith <greg(at)2ndquadrant(dot)com>
Cc: Anton Belyaev <anton(dot)belyaev(at)gmail(dot)com>, pgsql-general(at)postgresql(dot)org
Subject: Re: Huge iowait during checkpoint finish
Date: 2010-01-11 23:01:00
Message-ID: dcc563d11001111501w723256eel4b39f99e8a5e1131@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

On Mon, Jan 11, 2010 at 2:59 PM, Greg Smith <greg(at)2ndquadrant(dot)com> wrote:
> Scott Marlowe wrote:

> If you can shoehorn one more drive, you could run RAID-10 and get much
> better performance.
>
>
> And throwing drives at the problem may not help.  I've see a system with a
> 48 disk software RAID-10 that only got 100 TPS when running a commit-heavy
> test, because it didn't have any way to cache writes usefully for that
> purpose.

A 4 disk RAID-10 will be about 4 to 8 times faster than a RAID-5 of 3
disks. It won't be as fast as a good sized RAID-10 with HW caching,
but it would be a big improvement.

> If the old system had a write caching card, and the new one doesn't, that's
> certainly your most likely suspect for the source of the slowdown.  As for

Agreed. If the new machine is limited to 3 disks, and any one is big
enough to hold the db, I'd look at a two disk mirror with a hot spare
on a HW RAID controller with battery backed chat. If they can't get a
HW RAID controller than switching to SW RAID-1 would be a positive
step.

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Omar Mehmood 2010-01-11 23:02:18 replication from multiple "master" servers to a single read-only slave
Previous Message Greg Smith 2010-01-11 21:59:11 Re: Huge iowait during checkpoint finish