Report Optimization

[DiGzâ„¢747]

Hello guys,

We have a report that has an issue on its time performance.  I would like to know where do we start in improving the response time of a report in report studio.

Thanks alot,
DiGzâ,,¢747

[blom0344]

Hello guys,

We have a report that has an issue on its time performance.  I would like to know where do we start in improving the response time of a report in report studio.

Thanks alot,
DiGzâ,,¢747

The place to start is with running a trace with your DBA to check whether the SQL-statements compiled by Cognos that are send to the source database are really efficient. (This includes queries that are used to build prompt lists!!)
If these are sound and execute quickly , then look at the number of queries that are defined in the report  (as by default these are executed seqentially). (The more recent versions allow firing queries concurrently)
Check whether all defined calculated items are used in the report (creating them will mean they are added to the query , even if they are not used in an object)
Check if queries have been copied and contain redundant objects

Are you using extensive conditional formatting? Are you using lists with lots of columns that have there own complex expressions? Are you defining list reports that can also be generated through a much more efficient crosstab?
(I recently rebuild a few reports from list to crosstab-based and the performance went up from minutes to mere seconds)

[dssd]

I was reading your post here. I thought your point that crosstab could give better performance was quite interesting. I always thought that list or crosstab was just about presenting it differently and wouldnt have effect on performance. Any reason why the difference in performance?

[barrysaab]

I think the because crosstab stores values in summarized format ,where as list holds  data in detail format.Thanks

[blom0344]

Crosstab and list both represent data, there is no storage involved. 
dssd reacts on very old posting (2008) , but the lists involved were populated by individual dataitems (each with their own temporal calculation) and used conditional formatting as well.  The report itself was rather inefficient with dataitems extensively referencing other ones.