<?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: Measure network throughput in Operating System - HP-UX</title>
    <link>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604810#M532209</link>
    <description>maybe test with netperf?&lt;BR /&gt;&lt;A href="http://www.netperf.org/netperf/" target="_blank"&gt;http://www.netperf.org/netperf/&lt;/A&gt;</description>
    <pubDate>Mon, 22 Mar 2010 15:22:08 GMT</pubDate>
    <dc:creator>melvyn burnard</dc:creator>
    <dc:date>2010-03-22T15:22:08Z</dc:date>
    <item>
      <title>Measure network throughput</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604809#M532208</link>
      <description>Dears,&lt;BR /&gt;&lt;BR /&gt;We have one server that has more than one IP, one of the connections is configured on 10gb/sec B2B with another server, I want to test the throughput between them and verify the speed:&lt;BR /&gt;&lt;BR /&gt;I already done the following test:&lt;BR /&gt;[server1]# dd if=/dev/zero of=test bs=1024k count=1000&lt;BR /&gt;[server1]# timex scp test server2_IP:/var&lt;BR /&gt;&lt;BR /&gt;The results were normal, about 15MB/s and the 1GB took about 1 minute to finish, I think it should take lower time than what we have here.&lt;BR /&gt;&lt;BR /&gt;I used the server2 IP to ensure using the 10GB network.&lt;BR /&gt;Is my way correct? Any idea how to test? Any tools?</description>
      <pubDate>Mon, 22 Mar 2010 12:28:15 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604809#M532208</guid>
      <dc:creator>mhm</dc:creator>
      <dc:date>2010-03-22T12:28:15Z</dc:date>
    </item>
    <item>
      <title>Re: Measure network throughput</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604810#M532209</link>
      <description>maybe test with netperf?&lt;BR /&gt;&lt;A href="http://www.netperf.org/netperf/" target="_blank"&gt;http://www.netperf.org/netperf/&lt;/A&gt;</description>
      <pubDate>Mon, 22 Mar 2010 15:22:08 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604810#M532209</guid>
      <dc:creator>melvyn burnard</dc:creator>
      <dc:date>2010-03-22T15:22:08Z</dc:date>
    </item>
    <item>
      <title>Re: Measure network throughput</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604811#M532210</link>
      <description>Melvyn - your check is in the mail :)&lt;BR /&gt;&lt;BR /&gt;If your goal is to measure maximum throughput through a 10Gbit/s NIC then I would put netperf on both ends and run what I call a 1Mx64 TCP_STREAM test:&lt;BR /&gt;&lt;BR /&gt;netperf -H &lt;REMOTEIP&gt; -t TCP_STREAM -l 30 -- -s 1M -S 1M -m 64K&lt;BR /&gt;&lt;BR /&gt;BTW, which ones do you have AD385A (PCI-X 2.0) or AD386A (PCIe 1.1 x8) and are they in a correspondingly "beefy" I/O slot?  Also, these multiple IP addresses - are the IPs of the different NICs in the system in separate IP subnets, or the same IP subnet?  If the same, that will complicate making sure the test actually goes over the interface desired.&lt;/REMOTEIP&gt;</description>
      <pubDate>Mon, 22 Mar 2010 23:47:48 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604811#M532210</guid>
      <dc:creator>rick jones</dc:creator>
      <dc:date>2010-03-22T23:47:48Z</dc:date>
    </item>
    <item>
      <title>Re: Measure network throughput</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604812#M532211</link>
      <description>Thanks I'll consider that, they are HP AD385-60001 PCI-X cards, I don't know what "beefy" means? All the IP's falling in different sunets.&lt;BR /&gt;&lt;BR /&gt;BTW, how much bandwidth should I get from those 10G cards? Can anybody tell?</description>
      <pubDate>Tue, 23 Mar 2010 05:25:28 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604812#M532211</guid>
      <dc:creator>mhm</dc:creator>
      <dc:date>2010-03-23T05:25:28Z</dc:date>
    </item>
    <item>
      <title>Re: Measure network throughput</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604813#M532212</link>
      <description>"beefy" means if you have your NICs installed on slots with high bandwith, pcix or pce for example.</description>
      <pubDate>Tue, 23 Mar 2010 07:08:20 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604813#M532212</guid>
      <dc:creator>likid0</dc:creator>
      <dc:date>2010-03-23T07:08:20Z</dc:date>
    </item>
    <item>
      <title>Re: Measure network throughput</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604814#M532213</link>
      <description>Hi, &lt;BR /&gt;If you see major difference with scp throughput, it is possible that the scp may be throttled by its own design.   If so, consult the advice found in the HP-UX Secure Shell Getting Started Guide, Chapter 6 Enabling HP-UX Secure Shell to Take Advantage of High Speed Networks.  Link to the guide is &lt;A href="http://docs.hp.com/en/5992-4213/ch06.html" target="_blank"&gt;http://docs.hp.com/en/5992-4213/ch06.html&lt;/A&gt;&lt;BR /&gt;&lt;BR /&gt;Hope that helps,&lt;BR /&gt;Brian</description>
      <pubDate>Tue, 23 Mar 2010 11:34:20 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604814#M532213</guid>
      <dc:creator>Brian Hackley</dc:creator>
      <dc:date>2010-03-23T11:34:20Z</dc:date>
    </item>
    <item>
      <title>Re: Measure network throughput</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604815#M532214</link>
      <description>It has been quite some time since I did much of anything with the AD385A, but if you have it in a 266 MHz slot, the NIC Hardware should be capable of "link rate" in one direction.  The specifics of your situation, such as the performance of the CPUs you have, might dictate using more than one TCP connection to arrive at that performance.&lt;BR /&gt;&lt;BR /&gt;If you put the NIC into a 133 MHz slot, the best you can expect is about 7 Gbit/s tops.&lt;BR /&gt;&lt;BR /&gt;The default TCP window size in HP-UX 11iv3 (32768 bytes) is not sufficient to achieve maximum single-connection performance.  That is why my netperf example used "1M" (M == 1024^2) socket buffers - 1 MB should be more than sufficient - 256K might suffice, but I'm in the habit of using 1M.</description>
      <pubDate>Tue, 23 Mar 2010 15:36:34 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604815#M532214</guid>
      <dc:creator>rick jones</dc:creator>
      <dc:date>2010-03-23T15:36:34Z</dc:date>
    </item>
    <item>
      <title>Re: Measure network throughput</title>
      <link>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604816#M532215</link>
      <description>Thanks for your replies,&lt;BR /&gt;&lt;BR /&gt;Yes rick, the cards resides on 266 MHz slots, how can we use more than one TCP connection in such a thing?&lt;BR /&gt;&lt;BR /&gt;Do you suggest to change the window size?</description>
      <pubDate>Wed, 24 Mar 2010 06:57:16 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-hp-ux/measure-network-throughput/m-p/4604816#M532215</guid>
      <dc:creator>mhm</dc:creator>
      <dc:date>2010-03-24T06:57:16Z</dc:date>
    </item>
  </channel>
</rss>

