Re: Which processor runs better for Postgresql? - Mailing list pgsql-performance

From Dave Page
Subject Re: Which processor runs better for Postgresql?
Date
Msg-id E7F85A1B5FF8D44C8A1AF6885BC9A0E401388902@ratbert.vale-housing.co.uk
Whole thread Raw
In response to Which processor runs better for Postgresql?  (Steve Poe <steve.poe@gmail.com>)
Responses Re: Which processor runs better for Postgresql?  (Steve Poe <steve.poe@gmail.com>)
Re: Which processor runs better for Postgresql?  (Scott Marlowe <smarlowe@g2switchworks.com>)
List pgsql-performance

> -----Original Message-----
> From: pgsql-performance-owner@postgresql.org
> [mailto:pgsql-performance-owner@postgresql.org] On Behalf Of
> Joshua D. Drake
> Sent: 13 June 2006 20:44
> To: Scott Marlowe
> Cc: steve.poe@gmail.com; pgsql-performance@postgresql.org
> Subject: Re: [PERFORM] Which processor runs better for Postgresql?
>
> They probably do. They have probably standardized on Dell
> hardware. It
> is technically a dumb reason, but from a business standpoint
> it makes sense.

We use Dell here for those reasons these days, but thankfully are able
to suitably overspec everything to allow for significant growth and any
minor performance issues that they may have (we've never seen any
though). In Dell's defence we've never had a single problem with the
2850's or 1850's we're running which have all been rock solid. They also
have excellent OOB management in their DRAC cards - far better than that
in the slightly older Intel boxes we also run. That is a big selling
point for us.

> You can get extremely competitive quotes from IBM or HP as
> long as you
> say, "You are competing against Dell".

Dell beat them hands down in our experience - and yes, we have had
numerous quotes for HP and IBM kit, each of them knowing they are
competing against Dell.

> > Dells tend to perform poorly, period.  They choose low end
> parts (the
> > 2600's Serverworks chipset is widely regarded as one of the slowest
> > chipset for the P-IV there is.) and then mucking around
> with the BIOS of
> > the add in cards to make them somewhat stable with their
> dodgy hardware.
>
> I can confirm this.

And how old are the 2600's now?

Anyhoo, I'm not saying the current machines are excellent performers or
anything, but there are good business reasons to run them if you don't
need to squeeze out every last pony.

Regards, Dave.

pgsql-performance by date:

Previous
From: "Jim C. Nasby"
Date:
Subject: Re: 64-bit vs 32-bit performance ... backwards?
Next
From: PFC
Date:
Subject: Re: scaling up postgres