1 Reply Latest reply on Sep 19, 2017 3:10 PM by tramachandran

    PI Data Flow and Snapshot Rates

    jhd235

      Dear Community and PI Experts, https://pisquare.osisoft.com/videos/1343?comment-10797= first of all, big thank you for this great video explanation, much of the things is clear now with typical data flow; and there are a lot of the questions arises too:     1) can you, please point me to the right sections of the PI documentation on Data Flow, Snapshots, Event Queue;     2) are there some recomendations on fixing situations, where PI snapshot rate is keep raising and raising;     3) how we can see ongoing status of data flow (e.g. some status bar, dashboard or CLI command to see current snapshot rate)?     4) how we can identify source (e.g. PI interface), from where abnormal high rate data flow is coming? thank you Yermek

        • Re: PI Data Flow and Snapshot Rates
          tramachandran

          Hi Yermek, (1,2) KB00831 - Understanding event queue statistics and identifying problems discusses aspects of data flow and event queue statistics. If you suspect that you have a corrupted event queue on your PI Server then follow the steps on KB 3081OSI8. In the case of a corrupted buffer queue, follow the steps in KB00389 as mentioned in the article.

          (3) You can use either the "piartool -ss" command from the pi\adm command prompt(a), or check the ratio of the following two PI performance counters in the Windows Performance monitor(b):

          (a) "Snapshot Events" : "Events Sent to Queue"

          (b) PI Snapshot Subsystem counters: "Snapshot/sec : "Queued Events/sec"

          Ideally, this ratio should be between 10:1 to 100:1, meaning for every 10-100 events that come into the queue, one is sent to the archive. Minimally, it should be between 3:1 to 10:1, especially on systems with a larger point count (over 50,000 points) or systems with very high frequency data. 3226OSI8 - What are recommended Exception and Compression settings

          (4) 3223OSI8 - How to find and disconnect a rogue consumer or heavy hitter

          If you suspect certain points that have high data flow you can easily identify the point source from SMT ->Points ->Point Builder

          Let us know if you have further questions.

          1 of 1 people found this helpful