Creating "Seeding" Reports for Dynamic Cubes

[adam_mc]

I am experimenting with adding Dynamic Cube Packages to our new 10.2.1 environment.

To get the best use (for user experience), it seems I should create a number of "seeding" reports to build the cube (that will run off hours), so that the users get the best response time.

I have a Dynamic Cube setup that has the following dimensions/hierarchies:

- Region/District/Store (~5/10/350 entries in each level)
- Product Division/Department/Class (~20/500/3000 entries in each level)
- Time (Y/Q/M/W for potentially 3 rolling years)

The data is from a very large (IMO) Fact table that is at Store/Item/Week level.

I can run reports at higher levels (Region/District/Division/Week) and the seeding works great.
However, if I try and run a report at the lowest level (Store/Class/Week), the report never seems to finish or at least render (even though when I view in Toad the JDBC Thin Client session has ended).
And, the seeding does not seem to have occurred.

Are there any suggestions that anyone has as to how to accomplish this?
Or, are there any limitations to "size"?
I know an aggregate table in the DB would help, but unfortunately this is not an option at the moment.

Any thoughts would be greatly appreciated.
Thanks in advance,
Adam.

[navissar]

Try running the DQA to analyse what your cube does... There could be size limitations if your machine 's memory is isn't enough to keep all the aggregates And lowest level data cached.

[MFGF]

I'd also suggest that your "seeding" reports are redundant if you enable workload logging then run DQA to determine the most appropriate DB and in-memory aggregates, then apply these.

Cheers!

MF.