Please check the compression setting of your output PI Points. It's possible that outputs are being compressed out.
In addition to Steve's suggestion, if you have a PI Collective, I would double check that the data for the inputs is the same on all nodes.
It's important to distinguish that with the Analysis Data Reference, the AF client is performing the calculation, not the Analysis Service.
If our suggestions don't lead to a solution, can you post the analysis configuration and analysis service version?
Please find configuration details as requested. Analysis is writing data to PI points once we restart and it is taking time to catch up with the speed. We are receiving frequent updates every 10 sec and approx 20k calcs but looks like analysis service is not catching up with the speed.
PI AF analysis version : 2016 R2 ( 18.104.22.16859)
Analysis is almost 12 hours behind the actual trigger time and could notice updates in Data Archive but almost with a gap of 12-13 hours for every update we receive.
Lal Babu Shaik
2 of 2 people found this helpful
It sounds like you have a high latency issue with your Analysis Service. Is it at all possible to upgrade to 2018. The number of improvements since 2016 R2 in terms of performance are too long to list, but here are a few:
1) Number of Evaluation Threads exposed in 2017. You can use 4*P threads were P is the number of processors on the Analysis Server.
2) Better Statistics reporting in 2017, then again in 2017 R2 and again in 2018. Troubleshooting latency issues as of 2017 R2 is typically a matter of minutes compared to a few hours in 2016 R2.
3) Dependent groups are better managed so that poorly performing lower rank analyses don't affect unrelated upper rank analyses. Implemented in 2018.
4) Automatic cache time span, implemented in 2017.
5) UI enhanced to report a predictive statistic of performance in the Preview Result pane. This was implemented in 2018.
Once you have upgraded, if you are still experiencing latency issues, I would recommend reading Nitin Agarwal blog post:
It was written for the 2017 release, but the ideas presented are applicable to the versions 2017 R2 and 2018.
If it is not possible to upgrade, can you message me and send a statistics file ? You can get this file from PSE > Management Tab > in the right pane, under Operations, right click > View Service Statistics.
I can have a look a provide an answer here if the root cause is simple. If it's a little more complex, I'll open a support case for you and assist you via the case. We can then revert back to this blog post with the solution.
Thank you for detailed information on the same and will try to upgrade if it solves high volume updates.
We have raised a support case for performance issue and waiting for an update on request #884928 after log analysis.