Data limitations in Cube Building 1

[Vinstigator]

I have a need to build a Cube that contains over 30 million rows of data. Can anyone please advise if there are any limitations on the amount of data that either Transformer and/or Impromptu can handle? Please add any advise on best methods for addressing/processing such size.

I have been building cubes up to the 1 million row counts, and some have been requiring many hours to process. I am using Oracle on the back end to store the data. I know there are limits to the amounts of levels and categories, but how about data?

Thanks in advance,
Kimo

 

[Vinstigator]

One more question. Can anyone tell me of any advantages/disadvantages of running the server environment from a Windows Nt based to a Unix based environment? From a performance perspective. - Thanks!

 

[goldsmt]

Currently the data limits are 50 million rows, upto 500000 caegories.

I've built cubes upto 15 million, rather than go to directly to the data tables it was quicker (build time and devloping) to create an extract in one file containing all the fields required for the data model, then on a nightly basis the file is populated with only new data so to use incremental build process.

Tony Goldsmith