A little cube challenge I have - splitting a single cube into two

[eo]

SSAS 2005 over SQL Server 2005 DB'es

I have a single and comprehensive cube, which contains SCD's for many month ends. But due to a rapid increase in data volumes, we cannot run this cube on a daily basis any longer. So the decision was reached to have two copies of the same cube. One Current version, which shows a snapshot of the Facts for that day only, and a SCD version, which shows the monay month end positions (this will run over weekends).

Is there a way to leverage the same cube DSV, and create a seperate current version cube, but stipulate somehow to only look at the existing load date??

EO
Hertfordshire, England

 

[MDXer]

if the issue is the build time why not utilize partinioning to reduce the data volume being processed?

 

[eo]

we did consider this, but it proved to be quite conveluted as we required a SSIS package to manage running of the partitions, etc. I was hoping these was a way of creating a copy of the cube, and leveraging the existing DSV. Is there now way to speficy in a meaure group property to run the data over dataset A to Z, and on cube specify to run over only dataset A (for examp0le by way of a SQL query or some such)??

EO
Hertfordshire, England

 

[eo]

I never prrof read my message, so let me retype my last sentence...
"Is there not a way of specifying in a meaure group property on cube A to work over dataset A to Z, and for cube B to limit the dataset from the same fact table on the same DSV only to dataset A to L (for example by way of a SQL query or some such)?? "

EO
Hertfordshire, England

 

[MDXer]

in your partition definition. Remember all measure groups have 1 partition by default.

A comprehensive partitioning strategy is really the only way to accomplish what you need under either scenario. If you work through your needs the solution will require a bit more work but it is really not that difficult to implement automated partitioning.

The benefits that come with partinioning in my opinion far out weigh the time to implement it.