Re: Performance tuning in PostgreSQL

From: Dennis Gearon <gearond(at)cvc(dot)net>
To: "Daniel R(dot) Anderson" <dan(at)mathjunkies(dot)com>
Cc: Pgsql-General <pgsql-general(at)postgresql(dot)org>
Subject: Re: Performance tuning in PostgreSQL
Date: 2003-03-26 17:56:37
Message-ID: 3E81E9D5.3090908@cvc.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

In General, the rotational speed is higher on SCSCI disks, and this increases
the tranfer rate from the disc, which is the limitation for anything not in the
disk's cache. Given the same areal dinsity, a 15,000 SCSI drive will be 50%
faster in tranfer rate than a 10,000 IDE drive.

Daniel R. Anderson wrote:
> <snip>
>
>>SCSI is almost
>>always faster than IDE, all other things being equal (i.e. my 80 gig IDE
>>"mass storage" drives are way faster than a 2 Gig Ultra Wide SCSI drive
>>from 6 years ago would be, but any modern SCSI drive will kick the butt on
>>my IDE drives.
>
> </snip>
>
> That's not /entirely/ true. There was an article on slashdot a while
> back about what exactly the differences between IDE and SCSI are. IDE
> has pretty much almost caught up to SCSI in terms of everything except
> testing -- i.e. one of the reasons SCSI drives cost so much more is that
> they are each run through extensive individual tests to make sure
> they're not gonna break down 5 minutes out of the box.
>
> The only other difference, if I remember correctly, was the amount of
> drives you could put on the same cable. I'm going out on a limb here,
> but while ATA133 or whatever you're running /needs/ a single cable and
> controller to itself SCSI can put several drives on the same cable while
> maintaining speed.
>
> So the good news is that if money is tight you could probably justify an
> IDE raid, but if you really need that extra reliability SCSI might be
> the answer.

In response to

Responses

Browse pgsql-general by date

  From Date Subject
Next Message Camarao, Wagner (v) 2003-03-26 18:06:40 PostgreSQL
Previous Message Richard Huxton 2003-03-26 17:51:12 Re: Long term read-only records?