Performance: Cognos Cube x Sybase IQ

[paulista07]

Hello everyone!

I have reports that access the Sybase IQ database directly, but I'm thinking to create an intermediate Cognos Cube.

Thus, the reports would access the cube instead of Sybase IQ.

I'm having doubts about the performance. Would there be any improvement? Has anyone had any experience about this?

Thanks!

Fabiano Rodrigues

 

[blom0344]

Do you expect to reuse report designs from Sybase IQ for the Cognos cubes? I seriously doubt that will be the case. I would expect performance gains if you load the cubes with aggregated data, keep dimension set small and make sure that you have small sets of categories for each dimension level

Ties Blom

 

[paulista07]

Blom, thanks for your reply!

And if I load Sybase IQ tables with aggregated data instead of cognos cube?

My doubt is about the perfomance between cognos cube and sybase iq table with the same number of records. For example, the performance of a cube is better than the performance of a SQL Server table, but Sybase IQ is a database optimized for queries.

Compensate replace a table from Sybase IQ for a cube with the same amount of records?

In time, sorry for my poor English.

Thanks!

 

[blom0344]

You do not typically store a cube with non-aggregated data. If you do , then you will find that the actual size will be enormous. Each intersection between measure and dimension is accounted for, even if it contains no 'real' data. A cube is pretty useless if you need reporting down to the record level. It is primarely there for performing analysis on aggregated data.

You should therefor not think as a replacement for relational queries, but as an additional tool

Ties Blom