From: | Ow Mun Heng <Ow(dot)Mun(dot)Heng(at)wdc(dot)com> |
---|---|
To: | Magnus Hagander <magnus(at)hagander(dot)net> |
Cc: | Tomas Vondra <tv(at)fuzzy(dot)cz>, Ketema <ketema(at)gmail(dot)com>, pgsql-performance(at)postgresql(dot)org |
Subject: | Re: Hardware for PostgreSQL |
Date: | 2007-11-01 07:46:01 |
Message-ID: | 1193903161.9148.27.camel@neuromancer.home.net |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-performance |
On Thu, 2007-11-01 at 07:54 +0100, Magnus Hagander wrote:
> Ow Mun Heng wrote:
> > On Wed, 2007-10-31 at 22:58 +0100, Tomas Vondra wrote:
> >
> >> 2) separate the transaction log from the database
> >>
> >> It's mostly written, and it's the most valuable data you have. And in
> >> case you use PITR, this is the only thing that really needs to be
> >> backed up.
> >
> > My main DB datastore is in a raid1 array and the xlog is still
> > maintained in a single OS drive. Is this considered OK?
>
> Is your OS not RAIDed? I'd keep everything RAIDed one way or another -
> otherwise you are certain to get downtime if the disk fails.
Nope it's not raided. It's a very low end "server" running on IDE, max 4
drives. 1x80G system and 3x500G Raid1+1 hot spare
>
> Also, if you don't have a *dedicated* disk for the xlog (putting it on
> the OS disk doesn't make it dedicated), you miss out on most of the
> performance advantage of doing it. The advantage is in that the writes
> will be sequential so the disks don't have to seek, but if you have
> other access on the same disk, that's not true anymore.
As of right now, budget constraints is making me make do with that I've
got/(and it's not a whole lot)
>
> You're likely better off (performance-wise) putting it on the same disk
> as the database itself if that one has better RAID, for example.
I'm thinking along the lines of since nothing much writes to the OS
Disk, I should(keyword) be safe.
Thanks for the food for thought. Now.. time to find some dough to throw
around. :-)
From | Date | Subject | |
---|---|---|---|
Next Message | Magnus Hagander | 2007-11-01 07:52:21 | Re: Hardware for PostgreSQL |
Previous Message | Magnus Hagander | 2007-11-01 06:57:02 | Re: hardware and For PostgreSQL |