When should a TM1 cube be separated into multiple cub

[dfrench77]

Hi,

What are the reasons a TM1 cube should be separated into multiple cubes? I am not talking about a cube for different departments or functions. There is a cube that I am evaluating in which someone wants to separate by input, calculation, and output like for Cognos Enterprise Planning. This is an ordinary cube not extraordinarily large. I am looking for general thoughts.

Thanks,

dfrench77

[David Usherwood]

The simple answer is 'when the design demands it'. But that probably doesn't help you much .
My view is that you should have separate cubes when the dimensionality and content differs.
There is _no need_ to separate input, calculation and output, _or_ to separate by department - TM1's security copes with the latter perfectly well.

[lotsaram]

The only reason to separate cubes for input/output or by function or department or some other logical delineator etc for input is for performance reasons with large numbers of inputters where attempts to input data to the same cube at the same time cause performance issues for other inputters and readers alike (and readers likewise cause issues for writers). Otherwise I agree with David that security should be flexible enough to cope without need for additional cubes.

... Hopefully the parallel interaction feature 9.5.2 should finally do away with needing to separate out data entry for large planning applications.