I a trying to troubleshoot/improve our analysis service performance. Currently running PI 2018SP2. I have been looking over the statistics and lag regarding a specific analysis that is at the top of the list in the EvaluationStatisticsForCalculationGroups, that can be found within the analysis stats within PSE.

I am confused as to why this group even shows a large lag. I have attached a screen shot of the stats for this group.

The analyses in this group are scheduled to run daily at 2AM.

The group shows an AverageElapsedMilliseconds of 322378, approximately 5 minutes. The AverageTriggerMilliSeconds is 86400000 (1 day).

If I am understanding the statistics correctly, it only takes 5 minutes to run the calculations for that group. Since these calculations only run once a day, how can they possibly have lag or be “falling behind”, if they run once a day and take 5 mins to complete? It shows a lastlagmilliseconds value of approx. 45 minutes. Confused as to the 45 min lag on this group

Also how does that relate to the Maximum Latency (current lag), which is described as "Maximum amount of time (in milliseconds) between receiving a trigger event and executing the analysis and publishing results for that trigger event. ". Does Max Latency contain the latency of the group which currently has the highest lag? If so, our current lag is 1:09 (1 hour 9 mins, and the group at the top of the list in terms of lag only shows a lag of approx 45 mins. So that doesn't match up.

Hi Jonathon,

When we had this issue, we had to check each group to see the lag that matches the over all analysis lag data.

For some reason the top lag group doesn't always show the highest lagging calculations.

Once we find the matching group by checking all the calculation groups, we will need to check for the calculation(s) in the group that gives high latency.

Mostly the issue would be due to the calculation(s) being configured as event triggered and the tag triggering calculation would receive very frequent data. You can consider to update them as periodic calculations.

As you mentioned on improving the performance of analysis calculations, below Links would help you to improve the performance

Asset Analytics Best Practices - Part 1: Use Variables

Asset Analytics Best Practices - Part 2: Data Density and Data Pattern

Asset Analytics Best Practices - Part 3: Input Attributes

Asset Analytics Best Practices - Part 4: Analyses in Warning or Error

Asset Analytics Best Practices - Part 5: Scheduling