It's all about the answers!

Ask a question

Can we use JazzMon to track performance of specific actions in RTC


Michael Walker (99214201157) | asked Oct 09 '12, 7:33 p.m.
The JazzMon documentation mentions using the Metronome tool to see what web services are being used for certain actions (ex. checking in a file, opening a plan).  This is an attractive metric to track for our RTC instances using the JazzMon tool as we can see the average time to execute these actions, based on the web services called, for each of our instances.

However, reading the doc it seems this may not be accurate as some of the web services could be shared and reused between different actions.

Is there anyway to accurately accomplish this using JazzMon to track the average time for specific actions?

Accepted answer


permanent link
Dave Schlegel (331167) | answered Oct 10 '12, 11:47 a.m.
JAZZ DEVELOPER
Unfortunately as you pointed out many of the fine grained web services are shared by multiple user operations and may take more or less time based on which operation they are being used in.  Generally speaking, the server-based information can't reliably isolate which specific user operations are being performed by users.

That being said, it could still be a worthwhile experiment to use Metronome to determine how many of which web services are needed for a specific operation and then produce an estimate based on the average JazzMon response times plus network latency times for that many round trips. If there is much variation in amount of data returned or sent, say for large plan results, some correction for that might help (see the "Bytes Sent" value in both Metronome and JazzMon). The resulting estimate should be reasonably predictive but there are too many variables to be very accurate.

Also be aware that the JazzMon 1.3.0 release now supports computing response times based on sample intervals rather than relying on the cumulative times reported directly by the server. This will much more accurately reflect performance under the shifting load at different times of day and would be essential to any attempt to predict use case response times.

Michael Walker selected this answer as the correct answer

Comments
Michael Walker commented Oct 10 '12, 5:41 p.m.

Thanks Dave...appreciate the response.

Your answer


Register or to post your answer.


Dashboards and work items are no longer publicly available, so some links may be invalid. We now provide similar information through other means. Learn more here.