2 Replies Latest reply on Oct 16, 2009 10:39 AM by tdanner

    sub-millisecond ping

      Hello all,

      Thank you for taking the time to give this feature request more body. Orion is a valued tool within my company and we intend to make it better by giving Solarwinds feedback on our needs.

      As a propritairy market maker the infrastructure team is maintaining and monitoring a high performance network en server farm. In our line of work millisecond resolution is no longer sufficient and are therefore moving to a monitoring system capable of delivering sub-milisecond resolutions.

      Orion is not capable to provide this information in a sub-millisecond resolution. Therefore I writing to see if Solarwinds is considiring delivering this feature to us.

      Possibly your company is looking for the same feature and therefore I ask everyone who think this feature request is valuable to post a response. This allows Solarwinds to see the need to implement it and proiritize it accordingly. Thank you for taking the time for your response...

      As an example I can suggest looking at the implementation of hrPing developed by cFos Software GmbH. This hrPing software tool does dilever sub-milisecond response times using high performance counters provided by the Microsoft operating system.

      As an example:

      This is hrPING v2.39 by cFos Software GmbH -- http://www.cfos.de

      Using source IP address 10.5.3.155 to send packets
      Pinging XXXXXX [10.4.16.2]
      with 64 bytes data (92 bytes IP):

      Reply from 10.4.16.2: seq=0000 time=0.598ms TTL=125 ID=4b24
      Reply from 10.4.16.2: seq=0001 time=0.592ms TTL=125 ID=4b29
      Reply from 10.4.16.2: seq=0002 time=0.569ms TTL=125 ID=4b30
      Reply from 10.4.16.2: seq=0003 time=0.574ms TTL=125 ID=4b53

      This will result in 1 ms latency.

      This is hrPING v2.39 by cFos Software GmbH -- http://www.cfos.de

      Pinging XXXXX [10.5.3.121]
      with 64 bytes data (92 bytes IP):

      Reply from 10.5.3.121: seq=0000 time=0.392ms TTL=128 ID=58a8
      Reply from 10.5.3.121: seq=0001 time=0.311ms TTL=128 ID=58a9
      Reply from 10.5.3.121: seq=0002 time=0.297ms TTL=128 ID=58ac
      Reply from 10.5.3.121: seq=0003 time=0.311ms TTL=128 ID=58ad

      Statistics for aocs-nlam-001:
          Packets: sent=4, rcvd=4, error=0, lost=0 (0% loss) in 1.500301 sec
          RTTs of replies in ms: min/avg/max: 0.297 / 0.327 / 0.392

      This will result in 0 ms.

      Both the 1 and 0 ms response times are for us not acceptable.

      Thank you all,

      best regards,

      Martin Meuwese

        • Re: sub-millisecond ping

          After taking a good look it seems that some sub milliseconds are recorded. Values recorded are still not accurate becuase the majority is reported as 0,00 milliseconds and some are 0,30 milliseconds. The 0,30 milliseconds is the correct round-trip time so the questions is:

          Why does Orion record latencies of 0 milliseconds while having sub-millisecond capabilities.

            • Re: sub-millisecond ping
              tdanner

              The ResponseTime column in the Nodes table in the database is an integer and is recorded in milliseconds. The same is true of the columns in the historical response time tables ReponseTime_Detail, ResponseTime_Hourly, and ResponseTime_Daily.

              When we build a chart for the website, we take all the available response time samples in each interval and average them. On the custom chart page you can change the size of the sample interval.

              I think what you are seeing here is that most of your samples were recorded as 0 ms. Around 6:00, you had a few samples recorded as 1 ms. If those 1 ms samples were grouped in an interval with three 0 ms samples, you would get a 0,25 ms bar on the chart. If a 1 ms sample were grouped in an interval with four 0 ms samples, you would get a 0,20 ms bar on the chart.