Re: Performance large tables.

From: Greg Stark <gsstark(at)mit(dot)edu>
To: Vivek Khera <vivek(at)khera(dot)org>
Cc: PG-General General <pgsql-general(at)postgresql(dot)org>
Subject: Re: Performance large tables.
Date: 2005-12-13 19:09:21
Message-ID: 87slswervi.fsf@stark.xeocode.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general


Vivek Khera <vivek(at)khera(dot)org> writes:

> On Dec 13, 2005, at 2:49 AM, Franz(dot)Rasper(at)izb(dot)de wrote:
>
> > What is the performance difference between U320 15kRPM and U320 10kRPM ?
> > Does your RAID crontoller has some memory (e.g. 128 MB or 256 MB )
> > and something like memory backup write cache (like HP DL 380 server) ?
> > Do you use Intel or Opteron cpus ?
>
> The 15k drives have higher sustained throughput so theoretically they would be
> faster for sequential scans of data. I have no hard numbers about this,
> though. See my thread on choosing between them from last thursday.

Actually the 15k drives have only moderately higher throughput. The top of the
line 15k Maxtor has a maximum throughput of 98MB/s while my 3 year old 7200
rpm drive can get over 50MB/s. Newer 7200rpm drives would be better but they
don't seem to include throughput in their specs.

While that's not bad, the difference in seek time and rotational latency is
the main advantage of a faster drive. The seek time of a 7200rpm is about
8-9ms and the rotational latency about 4ms. The seek time of the to of the
line 15kRPM drive is about 3ms and the rotational latency 2ms.

In short while the throughput is less than doubled, the speed for random
access reads is almost tripled.

--
greg

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Karsten Hilbert 2005-12-13 19:21:20 Re: timestamp <-> ctime conversion question...
Previous Message Andrus 2005-12-13 18:49:46 Re: Toolkit for creating editable grid