5 Replies Latest reply on Dec 12, 2011 3:32 PM by RJKSolutions

    Retrieving from an interface node

      Can someone enlighten me somewhat.  


      If I use the OSIsoft SI Input Adapter I can retrieve data from the PI Server and from an interface node.  How does the PI Interface Node set up work, does the SI input adapter connect to the PIBuffss session and get updates from there before the data hits the PI Server?  What is the definition of a "PI Interface Node" from the SI input adapter perspective?  Does the adapter get data before exception/compression is applied by PIBuffss?

        • Re: Retrieving from an interface node

          Okay, found my answer.  At the moment we can only connect to a PI Server.  We have to wait for a new PI buffer Subsystem that has its own Update Subsystem and for new version of the Input Adapter before we can connect directly to an interface node.  Correct?


          Let's fast forward a couple of years and we have these.  Will the plan be for SI to be able to access data before exception processing or will it behave the same as if it is connected to a PI Server?

            • Re: Retrieving from an interface node

              Hi Rhys, Thank-you for the interest in StreamInsight.




              You are right about the current StreamInsight adapter support.  You can sign up for tag data through the Update Manager and receive all events (tag value data) coming from the interface. 




              There has been some talk about modifying the Buffer SubSystem to support signups like the Update Manager.  If this happens, we will definitely support signing up with BufSS directly.  In this scenario, I doubt StreamInsight will be able to directly influence data streams destined for the PI Server.  However, it would not be difficult to pass data streams under one set of tag names from the Interface into BufSS, process those streams using StreamInsight and emit streams with different tag names.  The original set of tags could be configured in a way that they would not be stored in the Archive and the second set oculd be configured to be stored.




              There has also been some discussions about providing StreamInsight support between the Interface and the Buffer.  This would allow people to perform analytics on raw data shortly after it is retrieved from the remote device.



                • Re: Retrieving from an interface node

                  Hey Erwin...thanks for the reply.


                  Funny you should mention the no archiving route, this is what is being done for testing.  Bunch of tags with no archiving are just providing the updates SI needs but the calculated results are then archived in the same PI System.  Would be awesome to do this directly from the source before it hits a PI System, but as you said it may come in the future.


                  Do you know if any people are using the Matrikon OPC SI Input Adapter to read data and then writing to a PI System, rather than using a PI OPC Interface?  (So you only ever see the analysed data streams, not the raw input.)  You almost have the option to provide many different compression algorithms against a PI Server without ever changing the core PI Server code base, especially if you can plug directly in to an Interface and switch off exception/compression for all tags.