2 Replies Latest reply on Mar 12, 2018 7:10 PM by jtokach

    ESXi 6.5 U1 Disk Performance

    jtokach Enthusiast



      I'm using WinSCP to copy a large file to a standalone ESXi 6.5U1 host datastore. WinSCP is uploading at a consistent rate of ~44 MB/s.


      ESXi Disk Performance is:

      Total Disk usage of ~87 MB/s

      Total Write 86.7 MB/s

      Total Read 0.3 MB/s


      I have verified the units are all in fact MB/s, and not a mixture of Mbps, etc. and have repeated the results with several files on two datastores.


      I'm not thrilled with the performance, but I'm more concerned about understanding why ESXi is reporting write usage of (strangely coincidental) 2x greater than the actual transfer rate.






      Message was edited by: Jim Updated to state ESXi 6.5U1

        • 1. Re: ESXi 6.5 U1 Disk Performance
          daphnissov Champion
          Community WarriorsvExpert

          Using SCP to copy a file to an ESXi host is not a good way to assess the performance capabilities of its storage. If you're really interested in its capabilities, you should use a performance benchmark tool like IOMeter to take a proper reading using a controllable methodology.

          • 2. Re: ESXi 6.5 U1 Disk Performance
            jtokach Enthusiast

            Thank you for the response! While I appreciate method and rigor, simply math can confirm this (assuming I'm converting properly):


            Windows to Windows 116 GB copied in 18 minutes = ((116/18)/60)*1000 = 110 MB/s

            Windows to ESXi 116 GB copied in 40 minutes = ((116/40)/60)*1000 = 48 MB/s


            That said, ESXi is stating a write rate double of the actual throughput and I don't know why. If there's a valid reason, then the performance would be expected. If not, then we have a performance problem as indicated by the data.