<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/" version="2.0">
  <channel>
    <title>topic Re: LUNs in Operating System - HP-UX</title>
    <link>https://community.hpe.com/t5/operating-system-hp-ux/luns/m-p/2881850#M101115</link>
    <description>I would choose RAID1/0.... In stark contrast to the above!!!!&lt;BR /&gt;&lt;BR /&gt;Why...&lt;BR /&gt;&lt;BR /&gt;Well OLTP generally means reading &amp;amp; writing from DB.  Furthur than this the writes are "generally" small.  updating customer details, placing orders etc.  Thus most writes will be RANDOM &amp;amp; SMALL.&lt;BR /&gt;&lt;BR /&gt;A RAID5 or RAID5DP/RAID6 stragtey will mean 1 small write within the stripe will cause the whole stripe to be read &amp;amp; the parity re-calculated then the small write will happen PLUS a parity update/write (or two if it is RAID5DP/RAID6).  This will cause very slow write performance as it will cause multiple reads &amp;amp; writes.&lt;BR /&gt;&lt;BR /&gt;IF you can protect against small writes, or are doing mostly reads then RAID5, RAID5DP or RAID6 will be fine.  &lt;BR /&gt;&lt;BR /&gt;Large writes on RAID5, RAID5DP or RAID6 are also OK (by large I mean writes that are of a similar size to the stripe, so 8kB stripe width, across 10 disks would mean ~ 80kB).  Reads are also OK (small or large) as they will not trigger any parity calculations or&lt;BR /&gt;&lt;BR /&gt;Basically does your OLTP database do small writes? If so avoid RAID5, RAID5DP or RAID6.&lt;BR /&gt;&lt;BR /&gt;One last thisng IF you have VA74xx you can use AutoRAID which SHOULD configure RAID5DP &amp;amp; RAID1/0 to make sure the small writes are less of a problem, but maximise you storage.&lt;BR /&gt;&lt;BR /&gt;Regards&lt;BR /&gt;&lt;BR /&gt;Tim</description>
    <pubDate>Tue, 14 Jan 2003 22:56:00 GMT</pubDate>
    <dc:creator>Tim D Fulford</dc:creator>
    <dc:date>2003-01-14T22:56:00Z</dc:date>
    <item>
      <title>LUNs</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/luns/m-p/2881848#M101113</link>
      <description>Hi,&lt;BR /&gt;&lt;BR /&gt;What is the best configuration of LUN splits&lt;BR /&gt;from a 73G drive either RAID1/0 or RAID5 for an OLTP database.&lt;BR /&gt;&lt;BR /&gt;Any ideas would be greatly appreciated.&lt;BR /&gt;&lt;BR /&gt;Thanks in advance,&lt;BR /&gt;Lanke&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt; &lt;BR /&gt;&lt;BR /&gt;</description>
      <pubDate>Tue, 14 Jan 2003 18:50:31 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/luns/m-p/2881848#M101113</guid>
      <dc:creator>lanke_1</dc:creator>
      <dc:date>2003-01-14T18:50:31Z</dc:date>
    </item>
    <item>
      <title>Re: LUNs</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/luns/m-p/2881849#M101114</link>
      <description>RAID 0/1&lt;BR /&gt;Simultaneous access to multiple mirrored pairs increases I/O performance. In general, the greater the number of mirrored pairs, the greater the increase in performance.&lt;BR /&gt;RAID 5&lt;BR /&gt;??? Provides high read throughput for small block-size requests (2 KB to 8 KB);&lt;BR /&gt;??? Write performance is limited by the need to perform four I/O operations per write request;&lt;BR /&gt;??? Because some I/O operations occur simultaneously, performance depends on the number of disks in the LUN. Additional disks may improve performance;&lt;BR /&gt;??? The I/O performance of RAID 5 benefits significantly from write caching.&lt;BR /&gt;RAID 0/1 is good for medium/large block size write transfers. RAID 5 is good for small/medium block size read transfers.&lt;BR /&gt;So in your situation I would choose RAID5.&lt;BR /&gt;You would forget about all these is you use Autoraid-based arrays (VA, EVA) because all these things are managed and optimized by their firmware&lt;BR /&gt;Eugeny</description>
      <pubDate>Tue, 14 Jan 2003 19:20:20 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/luns/m-p/2881849#M101114</guid>
      <dc:creator>Eugeny Brychkov</dc:creator>
      <dc:date>2003-01-14T19:20:20Z</dc:date>
    </item>
    <item>
      <title>Re: LUNs</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/luns/m-p/2881850#M101115</link>
      <description>I would choose RAID1/0.... In stark contrast to the above!!!!&lt;BR /&gt;&lt;BR /&gt;Why...&lt;BR /&gt;&lt;BR /&gt;Well OLTP generally means reading &amp;amp; writing from DB.  Furthur than this the writes are "generally" small.  updating customer details, placing orders etc.  Thus most writes will be RANDOM &amp;amp; SMALL.&lt;BR /&gt;&lt;BR /&gt;A RAID5 or RAID5DP/RAID6 stragtey will mean 1 small write within the stripe will cause the whole stripe to be read &amp;amp; the parity re-calculated then the small write will happen PLUS a parity update/write (or two if it is RAID5DP/RAID6).  This will cause very slow write performance as it will cause multiple reads &amp;amp; writes.&lt;BR /&gt;&lt;BR /&gt;IF you can protect against small writes, or are doing mostly reads then RAID5, RAID5DP or RAID6 will be fine.  &lt;BR /&gt;&lt;BR /&gt;Large writes on RAID5, RAID5DP or RAID6 are also OK (by large I mean writes that are of a similar size to the stripe, so 8kB stripe width, across 10 disks would mean ~ 80kB).  Reads are also OK (small or large) as they will not trigger any parity calculations or&lt;BR /&gt;&lt;BR /&gt;Basically does your OLTP database do small writes? If so avoid RAID5, RAID5DP or RAID6.&lt;BR /&gt;&lt;BR /&gt;One last thisng IF you have VA74xx you can use AutoRAID which SHOULD configure RAID5DP &amp;amp; RAID1/0 to make sure the small writes are less of a problem, but maximise you storage.&lt;BR /&gt;&lt;BR /&gt;Regards&lt;BR /&gt;&lt;BR /&gt;Tim</description>
      <pubDate>Tue, 14 Jan 2003 22:56:00 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/luns/m-p/2881850#M101115</guid>
      <dc:creator>Tim D Fulford</dc:creator>
      <dc:date>2003-01-14T22:56:00Z</dc:date>
    </item>
  </channel>
</rss>

