Announcement

Collapse
No announcement yet.

Some Windows Server 2016 vs. Linux Network Benchmarks

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #21
    Originally posted by phoronix View Post
    Phoronix: Some Windows Server 2016 vs. Linux Network Benchmarks

    A Phoronix Premium supporter recently requested some Windows vs. Linux networking performance benchmarks. That is being done as part of a larger comparison also featuring the popular BSDs, but for some initial measurements, here are some Netperf networking performance metrics on Microsoft Windows Server 2016 and various Linux distributions...

    http://www.phoronix.com/scan.php?pag...-Windows-Linux
    @Michael: On the opensuse factory mailinglist we had a short discussion about this benchmark. We are aware that this was only a preview, so you may take the following hints to improve the final benchmark, or the next one to come. If time and money permits, of course.
    I am citing Michal Kubeček [email protected] who has clearly much more knowledge of benchmarking or networks than me. Maybe you can find some valid points there:

    [Michael] goes to long details to explain what disks do server and client have but not a word about NICs; "Gigabit Ethernet" is not nearly enough - and that hidden "Realtek PCIe GBE Family + Microsoft ISATAP" (???) isn't much better. [...] No information about how were client and server connected [or] netfilter configuration (which can affect the latency quite a log) and other networking settings [...] the author was running netperf TCP_RR and UDP_RR test (i.e. one very specific aspect of networking performance) with [unknown] parameters [...]
    Some tests show variance so high that the results should have been discarded and tests repeated [...] netperf has "-I" and "-i" parameters to make things easier. He runs the same test with two different lengths [...] the results should be the same within a margin of statistic error; if they are not, it should be a warning sign that something was wrong. [...]

    One thing that really stands out is the UDP_RR test on Tumbleweed, in particular the 360s one. The variance itself [tells] the test went completely bonkers [...] even upper end of the indicated interval is still way below the results of the 60s test. [...] I quickly ran netperf UDP_RR between my two machines, one running 42.3 with 4.15.8 Kernel:stable kernel (i.e. essentially Tumbleweed), the other Tumbleweed with 4.16-rc4 kernel from Kernel:HEAD. The hardware is definitely worse than Mr. Larabel's and NICs aren't anything special either (on-board Realtek 8168evl and common consumer grade Intel (82541GI)). Both machines are running a KDE desktop (I only wanted to get some idea about the results) so I used "-I 99 -i 20,5" to make sure the results are not completely random. The result I got was... wait for it... 10359.48. Even with 1400 bytes of request/response size, I still get 3257.53. As the highest result in the Phoronix article is 918.92, I feel Mr. Larabel owes us some information about what he was actually testing and how.
    I just hope the next article isn't going to present TCP_STREAM results (measured on gigabit ethernet).
    Last edited by tomtomme; 03-16-2018, 08:49 AM.

    Comment


    • #22
      It's not surprising Linux is faster. Even if tests weren't done properly the biggest stock exchanges switched from windows to Linux. No one serious runs windows for serious tasks.

      Comment


      • #23
        Originally posted by Pawlerson View Post
        It's not surprising Linux is faster. Even if tests weren't done properly the biggest stock exchanges switched from windows to Linux. No one serious runs windows for serious tasks.
        Huh? Nearly every stock exchange in the world runs on the Tandem/Compaq/HP NonStop platform. Mission critical stuff like that doesn't run on Windows, or Linux, and never has.

        Comment


        • #24
          Originally posted by torsionbar28 View Post
          Huh? Nearly every stock exchange in the world runs on the Tandem/Compaq/HP NonStop platform. Mission critical stuff like that doesn't run on Windows, or Linux, and never has.
          https://www.pcworld.com/article/2380...ll_street.html

          Comment


          • #25
            Originally posted by torsionbar28 View Post
            Huh? Nearly every stock exchange in the world runs on the Tandem/Compaq/HP NonStop platform. Mission critical stuff like that doesn't run on Windows, or Linux, and never has.
            Really? NASDAQ uses Gentoo.

            The Chicago Mercantile Exchange, New York Mercantile Exchange, Frankfurt Stock Exchange, Eurex derivative exchange and Philippine Stock Exchange all use Red Hat.

            NYSE used to mostly be Solaris, but they are migrating towards linux. So is the LSE. Deutsche Borse is Linux, so is Xetra.

            HP-UX, AIX and Solaris have all been in decline.

            I doubt any exchange relies on a single platform, but to claim mission critical stuff doesn't run on linux is a bit outdated. We aren't in 1996.

            Comment


            • #26
              From the indirect quote above: "The result I got was... wait for it... 10359.48. Even with 1400 bytes of request/response size, I still get 3257.53. As the highest result in the Phoronix article is 918.92, I feel Mr. Larabel owes us some information about what he was actually testing and how."

              Its' surprisingly difficult to find any info online about what results to expect specifically from an out-of-the-box desktop system (which usually have firewall turned off, strangely), but apparently getting several thousand round trips in TCP_RR or UDP_RR is not uncommon in general (or, in my case, ping round trips within the local network, for example). Considering that except for one outlier, all results are in the 700-950 range, would suggest to me that any problem is not necessarily in the tests themselves, but potentially could be a bottleneck in the hardware/network configuration. Perhaps the network switch in the pic should have only two cables plugged in...

              Comment


              • #27
                For both netperf TCP_RR and UDP_RR, I'm getting a bit more than 4800 T/s, with default values, between a laptop and a cheap old computer, both running Ubuntu 17.10 with no special settings, and a probably average quality network switch in between.

                Comment

                Working...
                X