Re: 3 x PostgreSQL in cluster/redunant

From: Trent Shipley <tshipley(at)deru(dot)com>
To: pgsql-general(at)postgresql(dot)org
Subject: Re: 3 x PostgreSQL in cluster/redunant
Date: 2005-11-15 23:55:05
Message-ID: 200511151655.06172.tshipley@deru.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

On Tuesday 2005-11-15 13:06, Joshua D. Drake wrote:
> Michelle Konzack wrote:
> > Am 2005-11-14 16:54:41, schrieb Jim C. Nasby:
> >> On Mon, Nov 14, 2005 at 07:36:44PM +0100, Michelle Konzack wrote:
> >>> Hello *,
> >>>
> >>> I have three Sun Server where I have reserved on each Server a Raid-5
> >>> of 1 TByte for my PostgreSQL. The first PostgreSQL is up and running
> >>> with a database of 150 GByte.
> >>
> >> Keep in mind that databases and RAID5 generally don't mix very well.
> >
> > Can you explain me why?
>
> RAID 5 is very expensive for writes.
>
> > Unfortunatly the Controllers in the three SUN-Servers do not support
> > 300 GByte SCSI-Drives, so I have to continue with the Raid-5 of 16x
> > 76 GByte.
>
> Could you do RAID 10?
>
> Sincerely,
>
> Joshua D. Drake

I've seen books on tuning recommend RAID-5 into the low terrabyte range for
read-dominated databases (notably small data warehouse applications).

For very large multi-terrabye applications the suggestion is that RAID-50
along with streaming to and from stochastically accessed distributed storage
can partially hide the expense of writing to storage while bringing the money
cost of storage down considerably.

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Bruce Momjian 2005-11-16 00:51:39 Re: Does PG support updateable view?
Previous Message Bruce Momjian 2005-11-15 22:03:45 Re: Congratulations on 8.1