Netgate Discussion Forum
    • Categories
    • Recent
    • Tags
    • Popular
    • Users
    • Search
    • Register
    • Login

    OpenVPN Performance

    Scheduled Pinned Locked Moved OpenVPN
    6 Posts 3 Posters 1.0k Views
    Loading More Posts
    • Oldest to Newest
    • Newest to Oldest
    • Most Votes
    Reply
    • Reply as topic
    Log in to reply
    This topic has been deleted. Only users with topic management privileges can see it.
    • J
      JustConfused
      last edited by JustConfused

      HI All,
      I am having a performance issue with OpenVPN - Site to Site setup
      Site A - Server Side. Virtual pfsense, ESXi, 1Gb duplex connection. Server behind the pfsense running Windows Server and an FTP Site. Actual files to be served on a NAS behind the server, inaccessible to the VPN client. Symbolic link on server makes the files look and appear to be local.

      Site B - Client Side. Virtual pfsense, ESXi, 300/30 Virgin Internet. Vigin Hub in modem mode, EdgeRouter X acting as internet router. Then pfsense, then client (windows 10)

      Virtual pfsenses have 1GB memory and plenty of CPU. Neither are constrained.
      FTP is unencrypted

      I am getting a throughput of 5.5Mbps from server to client using FTP - which is somewhat less than ideal
      I have played with MTU (down to 1400) and increased / decreased the send/receive buffers - none of which have an effect.

      If I run a speedtest from site A - I get 900+/900+ - to be expected
      If I run a speedtest from site B - I get somewhat variable results depending on time of day - but fairly good (200+/20+)
      If I run an FTP to Site A from somewhere else I get 100Mb
      If I run an FTP to Site B from somewhere else I also get a decent speed - just more variable

      I am using the link to overnight transfer changed files from Site A to Site B and vice versa - but I am only getting 5.5Mb throughput which seems just wrong

      Any ideas for how to go about looking further? I am having difficulty interpreting these results

      Sean

      JKnottJ 1 Reply Last reply Reply Quote 0
      • JKnottJ
        JKnott @JustConfused
        last edited by

        @JustConfused

        One thing to bear in mind is that overall performance will be determined by the upload bandwidth at each end. The download bandwidth is irrelevant, as it's usually much higher than upload. What happens if you run iperf to measure bandwidth between sites? I see iperf v3 is now available for pfSense.

        PfSense running on Qotom mini PC
        i5 CPU, 4 GB memory, 32 GB SSD & 4 Intel Gb Ethernet ports.
        UniFi AC-Lite access point

        I haven't lost my mind. It's around here...somewhere...

        1 Reply Last reply Reply Quote 0
        • H
          HuskerDu
          last edited by

          What is the roundtrip delay between A and B ? A fairly high one (more than 150ms) with TCP size window can lead to "wrong" results (far less bandwidth than really available).

          Try running several FTP in parallel to highlight it.

          1 Reply Last reply Reply Quote 0
          • J
            JustConfused
            last edited by

            Between A&B, with no traffic is 29ms or so

            I am having problems getting iperf to work due to firewall rules preventing such traffic. I'll work on that

            JKnottJ 1 Reply Last reply Reply Quote 0
            • JKnottJ
              JKnott @JustConfused
              last edited by

              @JustConfused said in OpenVPN Performance:

              Between A&B, with no traffic is 29ms or so

              I am having problems getting iperf to work due to firewall rules preventing such traffic. I'll work on that

              You'll have to open up port 5201 for iperf v3. The graphical interface in pfSense says port 5001, which is for v2.

              PfSense running on Qotom mini PC
              i5 CPU, 4 GB memory, 32 GB SSD & 4 Intel Gb Ethernet ports.
              UniFi AC-Lite access point

              I haven't lost my mind. It's around here...somewhere...

              1 Reply Last reply Reply Quote 0
              • J
                JustConfused
                last edited by

                I am not ignoring this - I just broke everything quite badly - so am having to recover

                :-(

                Sean

                1 Reply Last reply Reply Quote 0
                • First post
                  Last post
                Copyright 2025 Rubicon Communications LLC (Netgate). All rights reserved.