Measuring true scheduled update performance in 7.8

[Steve Fernandez]

I am trying to get a much better handle on Spotfire's scheduled update performance. With hundreds of reports in our library it's difficult to know where the real issues are. From my analysis I can tell there are these times related to update:

Scheduled Start Time

Actual Start Time

End Time

Technically there's a 4th time which would be the time you set in the schedule itself. The time above is when it actually gets scheduled.

It seems like what I get in the Schedule & Routing view is the scheduled start time and a calculation of End Time - Scheduled Start Time to show the total time a report has been waiting to finish it's update. The problem is this also includes the time the report was spent waiting for an available update slot. I found I had to dig into the ACTIONLOG to find the delta between Actual Start Time and End Time to get the real length of time spend updating the report itself. Got all that figured out.

What I'm trying to do now is combined ACTIONLOG with the tables related to update schedules to get a full picture of what's going on with my schedules. But, I have not been able to figure out how to do this accurately. I can manually find the JOB_TASK entry that's responsible for what should be a specific instance of the scheduler updating a file in ACTIONLOG, but I can't find a way to tie them automatically. The only way I can see to try this would be to time based on time stamp down to the second. That's pretty janky though.

Has anyone attempted this

[Yann Schwartz]

I usually go to "Monitoring & Diagnostics" --> "instances" tab. There is a table there, with a column called "Average Loading Time". I consider that it's pretty accurate. Did you have a look there

[Steve Fernandez]

We cache most of our reports, so those "load" numbers are usually very small since that column is only showing how long it takes to load the file from the cache.  I'm looking for update times since we have hundreds of reports that get cached.

[Yann Schwartz]

We do the same, if by "we cache" you mean you ask the "scheduled updates" to load your report in memory.

 

Still this "Average Loading Time" displayed here shows me loading time up to 30 minutes. We have obviously no user waiting that long to open a report, so it's definitely the time it took to the scheduler to load the report in memory (exactly what you are looking for, I think)