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

Switching a package between OLAP and DMR source?

Started by GSpeare, 03 Jun 2009 06:43:59 AM

Previous topic - Next topic

GSpeare

Hi,

I'm new to Cognos and we are just starting to build our production environment. We have a star-schema data warehouse built in SQL and are deciding how to build out the semantic layer, which means choosing between DMR and OLAP. (We do not have OLAP cubes built yet but have Analysis Services and experience with it.) DMR seems like the more flexible choice but has performance concerns.

I am wondering if it's possible to build a package using one source and later switch the source to the other (i.e., DMR to OLAP or vice versa) without breaking reports? If we could experiment with DMR and then switch to OLAP if there are problems that would be ideal.


blom0344

Switching from DMR to Cognos cubes would mean pointing the package to the cube as datasource. No idea if it is possible, but what would be the point without having the cubes themselves? I wouldn't go the path you suggest as Cubes are essentially C7 technology whereas DMR seems to become really useful with C8.4 onwards..

GSpeare

I'll be happy to go with DMR if the performance holds up...there seems to be a lot of concern among the Cognos experts we have talked to.

blom0344

I think the real issue at steak is whether the combination of DMR and materialized views/indexed views/summary tables will solve the performance problem. As a DMR-based report triggers an SQL underneath , the cost of execution will likely cause performance issues compared with physical cubes.
However , if the SQL can be executed against pre-aggregates stored in the database (each vendor has it's own solution there, hence the multiple names), then the optimizer has the opportunity to use a much more efficient path to produce the required set.

Maintaining aggregates in the database is not a trifle and there are lots of restrictions , but I think it may ultimately beat having to refresh cubes..