IBM Support

How to Troubleshoot Performance with Cognos Analytics Reports Using Planning Analytics as a Datasource

Troubleshooting


Problem

You are encountering performance concerns with Cognos Analytics Reports that use Planning Analytics as a datasource, however do not know where to start.  The objective of this document is to help you better understand the components involved with reporting on data from Planning Analytics using Cognos Analytics, and the areas of the product that require focus.

Diagnosing The Problem

To diagnose a Cognos Analytics / Planning Analytics performance concern,  you must quantify the problem and provide specific details:  ​​​
  • Are all reports affected?  Or only one/some?
  • For any reports affected, how was the performance concern raised?
    • Did the report perform better previously?  What is the basis of comparison for the concern?
    • Is performance being compared to a different product/viewer, or the same product?
    • Is performance being compared to a different environment?  How do the environments/data compare?
    • Is performance being compared to a different version of the product? (PA or CA?)
    • Did more than one version change?  (were both PA and CA updated?)
      • If both CA and PA have changed, is it possible to isolate the problem by testing CA against the previous PA version?
      • Did the CA Datasource Connection Type change from Cognos TM1 to Planning Analytics?
    • Is the performance problem related to concurrent users?  What had been done to test/optimize your environment for concurrency?
    • Does the problem impact all users, or only some?  Is the problem related to geography/latency?
    • Does the problem include write-back to the TM1 Server?  (which in turn, can cause invalidation of cube cache / TM1 Rules that need to be refired)
Whether a performance concern impacts multiple reports or a single report, it is important to ensure consistency during the troubleshooting process:
  • Create a list of reports or specific actions, recording your performance observations ( # of seconds / minutes to perform X action )
  • Using the same reports or specific actions, record the performance observations from your test/development environment
  • Ensure all performance testing is consistent with the reports/actions identified.  Continuously record your observations / results
  • Clearly define the success criteria for your performance concern
  • How many seconds/minutes do you expect the report to perform within?
  • How did you determine this success criteria?
  • Clearly label any log files for the scenario(s) identified/captured
When working with IBM Support, the questions/focus points above need to be made very clear in your problem description. 
The following logs will be useful towards understanding where the problem may reside:  

Resolving The Problem

It is very difficult to provide one clear plan to resolve a performance concern, due to the various ways that a performance problem may manifest itself.  In order to resolve a performance problem with Cognos Analytics with a Planning Analytics Datasource, you need to determine where time is being spent during a request/query.

The execution of a typical Cognos Analytics report, looks like the following:
  • Cognos Analytics report is selected or scheduled to run
  • Cognos Analytics sends one or more requests to the TM1 Server
    • If using IBM Cognos TM1 datasource connection type, the requests are made using the TM1NGAPI
    • If using Planning Analytics datasource connection type, the requests are made using the TM1 REST API
  • TM1 Server processes the requests and provides a response to Cognos Analytics for each completed request
  • Cognos Analytics performs any additional local processing required to create the report output
  • Report is displayed (or saved) for the user
     
Many performance concerns with Cognos Analytics are related to the TM1 Server, and not the client tool (CA):
  • When doing performance testing/analysis, TM1 Server Thread output (TM1Top / Planning Analytics Administration) needs to be captured
  • Have a user run a report/perform an action in Cognos Analytics, and follow that user's thread on the TM1 Server using TM1Top/PAA
  • If most time is spent on the TM1 Server, can the problem be replicated without Cognos Analytics?
    • Test a view that you can replicate using the TM1 Architect Cube Viewer, does the problem persist?
      • If your Cognos Analytics report is somewhat complex, test a report that is more closely aligned with a Cube View from Architect
      • You may not have an issue with Cognos Analytics at all, but a TM1 Server Performance problem and perhaps, an inefficiently designed TM1 Model/Report

If the majority of time spent for your report/action does not appear to be on the TM1 Server, the time may be spent with Cognos Analytics:
  • When doing performance testing/analysis, monitor the CPU/Memory for the Report Server's JAVA process
  • After the TM1 Server has finished processing the query, do you observe high CPU on the Cognos Analytics Report Server?
  • If CPU/Memory is high on the Report Server, you may need to better tune (or distribute) the Cognos Analytics environment
  • Performing all of the steps above will allow you to better understand where the problem you are troubleshooting resides.  
 
To summarize the above - it is critical that you:
  • Document all steps/descriptions/success criteria of performance concerns consistently
  • Ensure you have performed load/stress testing prior to any Production implementation
  • Test a number of scenarios to determine which component your problem truly resides in
  • Capture the suggested logging in order to paint a complete picture of the user experience
     
Important Notes:
Troubleshooting performance concerns is very much a collaborative process.  When providing logs for a Performance concern, you must provide details on who was performing what action, what time the action occurred, and the outcome/observations from that testing.  Successfully troubleshooting a performance concern is more about the documentation/details captured, and less about the technical.
 
In addition to all items above, if you have a problem that is intermittent or only impacts a handful of users - the same set of logs/details should be provided from a scenario which does not demonstrate the performance problem - as this comparison will allow for a better understanding of differences.
 
Lastly, if you have a scenario which can be replicated on demand - it may be best to provide everything required to replicate that scenario to IBM Support in order to further expedite resolution to the problem.  This would include:
  • TM1 Data Directory
  • TM1S.cfg file
  • Cognos Analytics Content Deployment (with Package and Report(s))
  • Steps required to recreate

Document Location

Worldwide

[{"Business Unit":{"code":"BU059","label":"IBM Software w\/o TPS"},"Product":{"code":"SSCTEW","label":"IBM Planning Analytics Local"},"Component":"","Platform":[{"code":"PF033","label":"Windows"}],"Version":"All Versions","Edition":"","Line of Business":{"code":"LOB10","label":"Data and AI"}}]

Document Information

Modified date:
21 September 2021

UID

ibm16114046

Page Feedback