Re: Better Hardware, worst Results - Mailing list pgsql-performance

From Tom Lane
Subject Re: Better Hardware, worst Results
Date
Msg-id 4847.1099604773@sss.pgh.pa.us
Whole thread Raw
In response to Better Hardware, worst Results  (Alvaro Nunes Melo <al_nunes@atua.com.br>)
List pgsql-performance
Alvaro Nunes Melo <al_nunes@atua.com.br> writes:
> I have a very tricky situation here. A client bought a Dell dual-machine
> to be used as Database Server, and we have a cheaper machine used in
> development. With identical databases, configuration parameters and
> running the same query, our machine is almost 3x faster.

> ==> Dell PowerEdge:
> HD: SCSI

> ==> Other machine:
> HD: IDE

I'll bet a nickel that the IDE drive is lying about write completion,
thereby gaining a significant performance boost at the cost of probable
data corruption during a power failure.  SCSI drives generally tell the
truth about this, but consumer-grade IDE gear is usually configured to
lie.

            regards, tom lane

pgsql-performance by date:

Previous
From: Rod Taylor
Date:
Subject: Re: Better Hardware, worst Results
Next
From: Pierre-Frédéric Caillaud
Date:
Subject: Re: Restricting Postgres