pg.ddx.io  pgsql-performance@postgresql.org mailing list archive  
help / color / mirror / Atom feed
From: Robert Schnabel <schnabelr@missouri.edu>
To: Scott Carey <scott@richrelevance.com>
Cc: Alan Hodgson <ahodgson@simkin.ca>
Cc: pgsql-performance@postgresql.org <pgsql-performance@postgresql.org>
Subject: Re: How to best use 32 15k.7 300GB drives?
Date: Fri, 28 Jan 2011 15:33:28 -0600
Message-ID: <4D433628.6010209@missouri.edu> (raw)
In-Reply-To: <C9683F12.1E3D5%scott@richrelevance.com>
References: <C9683F12.1E3D5%scott@richrelevance.com>

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#ffffff" text="#000000">
    <br>
    On 1/28/2011 11:44 AM, Scott Carey wrote:
    <blockquote cite="mid:C9683F12.1E3D5%25scott@richrelevance.com"
      type="cite">
      <div><br>
      </div>
      <div><br>
      </div>
      <span id="OLK_SRC_BODY_SECTION">
        <div>
          <div>On 1/27/11 4:11 PM, "Alan Hodgson" &lt;<a
              moz-do-not-send="true" href="mailto:ahodgson@simkin.ca">ahodgson@simkin.ca</a>&gt;
            wrote:</div>
        </div>
        <div><br>
        </div>
        <blockquote id="MAC_OUTLOOK_ATTRIBUTION_BLOCKQUOTE"
          style="padding: 0pt 0pt 0pt 5px; margin: 0pt 0pt 0pt 5px;">
          <div>
            <style type="text/css">
p, li { white-space: pre-wrap; }
</style>
            <div style="font-family: 'Ubuntu'; font-size: 11pt;
              font-weight: 400; font-style: normal;">
              <p style="margin: 0px; text-indent: 0px;">On January 27,
                2011, Robert Schnabel &lt;<a moz-do-not-send="true"
                  href="mailto:schnabelr@missouri.edu">schnabelr@missouri.edu</a>&gt;
                wrote:</p>
              <p style="margin: 0px; text-indent: 0px;">&gt; So my
                questions are 1) am I'm crazy for doing this, 2) would
                you change</p>
              <p style="margin: 0px; text-indent: 0px;">&gt; anything
                and 3) is it acceptable to put the xlog &amp; wal (and
                perhaps tmp</p>
              <p style="margin: 0px; text-indent: 0px;">&gt; filespace)
                on a different controller than everything else? Please
                keep</p>
              <p style="margin: 0px; text-indent: 0px;">&gt; in mind I'm
                a geneticist who happens to know a little bit about</p>
              <p style="margin: 0px; text-indent: 0px;">&gt;
                bioinformatics and not the reverse. :-)</p>
              <p style="margin: 0px; text-indent: 0px;">&gt; </p>
              <p style="margin: 0px; text-indent: 0px;">Putting the WAL
                on a second controller does help, if you're write-heavy.</p>
              <p style="margin: 0px; text-indent: 0px;">I tried
                separating indexes and data once on one server and
                didn't really notice that it helped much. Managing the
                space was problematic. I would suggest putting those
                together on a single RAID-10 of all the 300GB drives
                (minus a spare). It will probably outperform separate
                arrays most of the time, and be much easier to manage.</p>
            </div>
          </div>
        </blockquote>
      </span>
      <div><br>
      </div>
      <div>If you go this route, I suggest two equally sized RAID 10's
        on different controllers fir index + data, with software raid-0
        on top of that. &nbsp;RAID 10 will max out a controller after 6 to 10
        drives, usually. &nbsp;Using the OS RAID 0 to aggregate the
        throughput of two controllers works great.</div>
      <div><br>
      </div>
      <div>WAL only has to be a little bit faster than your network in
        most cases. &nbsp;I've never seen it be a bottleneck on large bulk
        loads if it is on its own controller with 120MB/sec write
        throughput. &nbsp;I suppose a bulk load from COPY might stress it a
        bit more, but CPU ends up the bottleneck in postgres once you
        have I/O hardware this capable.</div>
      <div><br>
      </div>
    </blockquote>
    Do you mean 14 drives in one box as RAID10's on one controller, then
    14 drives in the other box on a second controller, then software
    RAID0 each of the two RAID10's together essentially as a single 4 TB
    array?&nbsp; Would you still recommend doing this with Windows?<br>
    Bob<br>
    <br>
    <br>
    <br>
  </body>
</html>



view thread (24+ messages)  latest in thread

Message-ID: <4D433628.6010209@missouri.edu>
Permalink:  ../4D433628.6010209@missouri.edu/
Also on:    postgresql.org/message-id/4D433628.6010209@missouri.edu

 · 

reply

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Reply to all the recipients using the --to and --cc options:
  reply via email

  To: pgsql-performance@postgresql.org
  Cc: schnabelr@missouri.edu, scott@richrelevance.com, ahodgson@simkin.ca
  Subject: Re: How to best use 32 15k.7 300GB drives?
  In-Reply-To: <4D433628.6010209@missouri.edu>

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox