If you are unable to create a new account, please email support@bspsoftware.com

 

News:

MetaManager - Administrative Tools for IBM Cognos
Pricing starting at $2,100
Download Now    Learn More

Main Menu

Performance

Started by dssd, 30 Aug 2013 09:42:42 AM

Previous topic - Next topic

dssd

Are list faster than crosstab and charts everything else remaining same?

BigChris

Potentially - a list by default will only bring back a page of data at a time. If the crosstab is summarising lots of data so that it all comes back on a single page, for example, it'll obviously take longer. Ditto with charts.

blom0344

In my early Cognos days I ran into a set of reports (full client BO) that contained massive crosstabs. The idea was to rebuild reports in Cognos (including calculations) , but we found that the limit for very large crosstabs is easily reached with webbased reports. I never noticed this with simple list layouts though, no matter how large the list performance stays acceptable..

CognosPaul

The reason crosstabs take longer than lists is fairly simple.

Let's take a simple scenario: You have a simple list of a two thousand rows, maybe one hundred locations * 20 products, and a measure. The same query is feeding a crosstab with locations on the rows, products in the column.

To begin with, Cognos won't be able to use the same query on both data containers. (Unless you're using DQM, but that's a whole 'nother horse of a different can of worms.)

The list will work with a cursor, if database supported, so it will begin returning data immediately. As the page is generating, the query will still be retrieving data. When you hit the page down, it will simply move the cursor down n rows.

On the other hand, a crosstab needs to process the entire query before it can render. Additionally, crosstabs aren't generally supported by databases, so Cognos needs to do some local processing to pivot the data, adding to the execution time.

This is actually one of the reasons that I try to avoid using crosstabs off relational sources.