<?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: EVA4400 performance issue in HPE EVA Storage</title>
    <link>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260335#M59946</link>
    <description>Hi,&lt;BR /&gt;&lt;BR /&gt;Why do a IOtest which adds software striping, to a EVA diskarray that allready, afaik, does hardware striping of all IO by default ? (and software striping over hardware striping normally worsens the performance) Remove the software striping parameter and see what the results are then.&lt;BR /&gt;&lt;BR /&gt;Greetz,&lt;BR /&gt;Chris</description>
    <pubDate>Tue, 26 Oct 2010 20:12:08 GMT</pubDate>
    <dc:creator>chris huys_4</dc:creator>
    <dc:date>2010-10-26T20:12:08Z</dc:date>
    <item>
      <title>EVA4400 performance issue</title>
      <link>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260333#M59944</link>
      <description>&lt;!--!*#--&gt;On EVA4400 with 2 disk enclosure and 300GB 15K FC Drives, all 24 disk is in one Disk Group. I created a Vdisk 100Gb VRAID1, present it to Server 2008 x64 (Qlogiq HBA) and performance test was&lt;BR /&gt;&lt;BR /&gt;sqlio -kW -t3 -s60 -b4&lt;BR /&gt;sqlio v1.5.SG&lt;BR /&gt;3 threads writing for 60 secs to file testfile.dat using 4KB IOs over 256KB stripes with 64 IOs per run initialization done&lt;BR /&gt;CUMULATIVE DATA:&lt;BR /&gt;throughput metrics:&lt;BR /&gt;IOs/sec:  9488.56&lt;BR /&gt;MBs/sec:    37.06 &lt;BR /&gt;&lt;BR /&gt;Then I create a disk 350Gb, RAID1 on and present it to the same host, speed is twice different!&lt;BR /&gt;D:\SQLIO&amp;gt;sqlio -kW -t3 -s60 -b4&lt;BR /&gt;sqlio v1.5.SG&lt;BR /&gt;3 threads writing for 60 secs to file testfile.dat using 4KB IOs over 256KB stripes with 64 IOs per run initialization done&lt;BR /&gt;CUMULATIVE DATA:&lt;BR /&gt;throughput metrics:&lt;BR /&gt;IOs/sec:  4675.58&lt;BR /&gt;MBs/sec:    18.26&lt;BR /&gt;&lt;BR /&gt;Why its happens? VDisk from same storage in the same Disk Group has different performance?&lt;BR /&gt;</description>
      <pubDate>Tue, 26 Oct 2010 17:34:50 GMT</pubDate>
      <guid>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260333#M59944</guid>
      <dc:creator>Arman Obosyan</dc:creator>
      <dc:date>2010-10-26T17:34:50Z</dc:date>
    </item>
    <item>
      <title>Re: EVA4400 performance issue</title>
      <link>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260334#M59945</link>
      <description>Hello,&lt;BR /&gt;really stange, as you use the same disk/raid it shoudnt be different.&lt;BR /&gt;Try other parameters/ benchmarks?&lt;BR /&gt;&lt;BR /&gt;Jan</description>
      <pubDate>Tue, 26 Oct 2010 19:30:12 GMT</pubDate>
      <guid>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260334#M59945</guid>
      <dc:creator>Jan Soska</dc:creator>
      <dc:date>2010-10-26T19:30:12Z</dc:date>
    </item>
    <item>
      <title>Re: EVA4400 performance issue</title>
      <link>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260335#M59946</link>
      <description>Hi,&lt;BR /&gt;&lt;BR /&gt;Why do a IOtest which adds software striping, to a EVA diskarray that allready, afaik, does hardware striping of all IO by default ? (and software striping over hardware striping normally worsens the performance) Remove the software striping parameter and see what the results are then.&lt;BR /&gt;&lt;BR /&gt;Greetz,&lt;BR /&gt;Chris</description>
      <pubDate>Tue, 26 Oct 2010 20:12:08 GMT</pubDate>
      <guid>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260335#M59946</guid>
      <dc:creator>chris huys_4</dc:creator>
      <dc:date>2010-10-26T20:12:08Z</dc:date>
    </item>
    <item>
      <title>Re: EVA4400 performance issue</title>
      <link>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260336#M59947</link>
      <description>Jan Soska,&lt;BR /&gt;Yes for example test it 64k size (sqlio -kW -t3 -s60 -b64)&lt;BR /&gt;And test results are same on LUNs&lt;BR /&gt;&lt;BR /&gt;100Gb Raid1 VDISK&lt;BR /&gt;L:\SQLIO&amp;gt;sqlio -kW -t3 -s60 -b64&lt;BR /&gt;sqlio v1.5.SG&lt;BR /&gt;3 threads writing for 60 secs to file testfile.dat using 64KB IOs over 4096KB stripes with 64 IOs &lt;BR /&gt;initialization done&lt;BR /&gt;CUMULATIVE DATA:&lt;BR /&gt;throughput metrics:&lt;BR /&gt;IOs/sec:  3540.73&lt;BR /&gt;MBs/sec:   221.29&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt;350Gb Raid1 VDISK&lt;BR /&gt;D:\SQLIO&amp;gt;sqlio -kW -t3 -s60 -b64&lt;BR /&gt;sqlio v1.5.SG&lt;BR /&gt;3 threads writing for 60 secs to file testfile.dat using 64KB IOs over 4096KB stripes with 64 IOs &lt;BR /&gt;initialization done&lt;BR /&gt;CUMULATIVE DATA:&lt;BR /&gt;throughput metrics:&lt;BR /&gt;IOs/sec:  3744.12&lt;BR /&gt;MBs/sec:   234.00&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt;But why is so different wane I use 4k size? May be its depends on vdisk size?&lt;BR /&gt;&lt;BR /&gt;Chris Huys,&lt;BR /&gt;Remove the software striping parameter? Where or how? I'm not using any stripe</description>
      <pubDate>Wed, 27 Oct 2010 04:39:56 GMT</pubDate>
      <guid>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260336#M59947</guid>
      <dc:creator>Arman Obosyan</dc:creator>
      <dc:date>2010-10-27T04:39:56Z</dc:date>
    </item>
    <item>
      <title>Re: EVA4400 performance issue</title>
      <link>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260337#M59948</link>
      <description>hello,&lt;BR /&gt;I do not know sqlio software, so I can not help with stripping options.&lt;BR /&gt;&lt;BR /&gt;Regarding block size - different block size results in very different results. Smallar block usually produce small r/w perf - as they require more IOs. Each drive is cabable only of limited amounth of IO - in case of 15K FC drives let say 170 IOps per drive. If you use bigger block - you need less IO and get better w/r performance.&lt;BR /&gt;But remember - this works only in case of large block used by application - which is usually not true (maybe in video editing it is ok). So you have to try block size used in your app, or tune the application how it works with different block size.&lt;BR /&gt;If I am not wrong, EVA's use 128K block natively.&lt;BR /&gt;&lt;BR /&gt;Jan</description>
      <pubDate>Wed, 27 Oct 2010 05:24:23 GMT</pubDate>
      <guid>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260337#M59948</guid>
      <dc:creator>Jan Soska</dc:creator>
      <dc:date>2010-10-27T05:24:23Z</dc:date>
    </item>
    <item>
      <title>Re: EVA4400 performance issue</title>
      <link>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260338#M59949</link>
      <description>Disk are prepared for SQL, that why I decide to use sqlio, is Microsoft tool, SQLIO Disk Subsystem Benchmark Tool&lt;BR /&gt;&lt;A href="http://www.microsoft.com/downloads/en/details.aspx?familyid=9A8B005B-84E4-4F24-8D65-CB53442D9E19&amp;amp;displaylang=en" target="_blank"&gt;http://www.microsoft.com/downloads/en/details.aspx?familyid=9A8B005B-84E4-4F24-8D65-CB53442D9E19&amp;amp;displaylang=en&lt;/A&gt;&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt;</description>
      <pubDate>Wed, 27 Oct 2010 06:08:45 GMT</pubDate>
      <guid>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260338#M59949</guid>
      <dc:creator>Arman Obosyan</dc:creator>
      <dc:date>2010-10-27T06:08:45Z</dc:date>
    </item>
    <item>
      <title>Re: EVA4400 performance issue</title>
      <link>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260339#M59950</link>
      <description>Hi,&lt;BR /&gt;&lt;BR /&gt;&amp;gt; D:\SQLIO&amp;gt;sqlio -kW -t3 -s60 -b4&lt;BR /&gt;&amp;gt; sqlio v1.5.SG&lt;BR /&gt;&amp;gt; 3 threads writing for 60 secs to file &lt;BR /&gt;&amp;gt; testfile.dat using 4KB IOs "over 256KB &lt;BR /&gt;&amp;gt; stripes" with 64 IOs per run initialization&lt;BR /&gt;&amp;gt; done&lt;BR /&gt;&lt;BR /&gt;The "over 256KB stripes" suggest the sqlio tool is emulating striping with the "given sqlio command options"..&lt;BR /&gt;&lt;BR /&gt;While f.e. from the following url :&lt;BR /&gt;&lt;A href="http://communities.vmware.com/thread/90464" target="_blank"&gt;http://communities.vmware.com/thread/90464&lt;/A&gt;&lt;BR /&gt;&lt;BR /&gt;, the following sqlio command doesnt do "striping"..&lt;BR /&gt;&lt;BR /&gt;C:\Program Files\SQLIO&amp;gt;"C:\Program Files\SQLIO\sqlio.exe" -frandom -i1 &lt;BR /&gt;sqlio v1.5.SG &lt;BR /&gt;1 thread reading for 30 secs from file testfile.dat &lt;BR /&gt;using 2KB random IOs &lt;BR /&gt;using current size: 8 MB for file: testfile.dat &lt;BR /&gt;initialization done &lt;BR /&gt;&lt;BR /&gt;Also I suppose, that the eva4400 should be giving results, comparable, with the following sites results for a eva6100..&lt;BR /&gt;&lt;BR /&gt;&lt;A href="http://blogs.mssqltips.com/blogs/chadboyd/archive/2008/03/16/ssd-and-sql-sqlio-performance.aspx" target="_blank"&gt;http://blogs.mssqltips.com/blogs/chadboyd/archive/2008/03/16/ssd-and-sql-sqlio-performance.aspx&lt;/A&gt;&lt;BR /&gt;&lt;BR /&gt;Greetz,&lt;BR /&gt;Chris</description>
      <pubDate>Wed, 27 Oct 2010 21:56:41 GMT</pubDate>
      <guid>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260339#M59950</guid>
      <dc:creator>chris huys_4</dc:creator>
      <dc:date>2010-10-27T21:56:41Z</dc:date>
    </item>
    <item>
      <title>Re: EVA4400 performance issue</title>
      <link>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260340#M59951</link>
      <description>Hello, another good testing software is HF Tune Pro, trial version works for 15 days, you can specify block size etc...&lt;BR /&gt;&lt;BR /&gt;Jan</description>
      <pubDate>Thu, 28 Oct 2010 10:30:02 GMT</pubDate>
      <guid>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260340#M59951</guid>
      <dc:creator>Jan Soska</dc:creator>
      <dc:date>2010-10-28T10:30:02Z</dc:date>
    </item>
    <item>
      <title>Re: EVA4400 performance issue</title>
      <link>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260341#M59952</link>
      <description>thanks to all.</description>
      <pubDate>Fri, 29 Oct 2010 14:13:53 GMT</pubDate>
      <guid>https://community.hpe.com/t5/hpe-eva-storage/eva4400-performance-issue/m-p/5260341#M59952</guid>
      <dc:creator>Arman Obosyan</dc:creator>
      <dc:date>2010-10-29T14:13:53Z</dc:date>
    </item>
  </channel>
</rss>

