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

validating a dimension error

Started by maxchuie, 23 Jul 2015 10:53:38 AM

Previous topic - Next topic

maxchuie

I am building dimensions and validating them i keep getting this error just as 
The data source is FWM model Package in Dynamic Mode source is SQL Server.. I have been informed that the dm is star schema.


The dimension (NAME) is a snowflake dimension. Queries using snowflake dimensions perform slower than queries using dimension tables because additional table joins are required in generated sql

I don't know what more I can do help improve performance in fwm and cube designer

bus_pass_man

It's a performance issue, not an error.  Errors will only show up in the issues tab.  You can get warnings in the issues tab too.   

Performance issues are a different thing from errors or warnings.

"Cognos Cube Designer also attempts to identify potential performance issues with the cube.
Performance issues are listed in the Performance tab. They identify model and database
issues that could cause slower queries or greater memory usage. You should consider the
issues identified, but if they are not relevant for your application, you can ignore them."

I direct you to this section in the new redbook.  One topic in it is about snowflakes.

11.10.3 Optimizing the cube model

Deep in the boring-est bowels of query optimization you will learn that if you need to do a join between tables in the query the performance will be slower than if it is getting results from one big dimension table.   

Depending on the situation, you may want or need to deal with the trade offs of the performance issue situations or simply bite the bullet. As a modeler, you don't necessarily need to know all the details about them but knowing that there is such cases that need to be considered and to be able to talk about them and bring them onto the agenda is important.

No matter what, snowflake is an industry standard term and a fairly elementary one at that. I believe that something you should consider investing more time in would be reading Kimball and similar books (I like Adamson's star schema) and other educational activities.