The reason this is occurring is due to the fact when you incrementally
process a dimension, a shadow copy of the dimension is created and stored in
memory along with the original. This is so users can continue to query the
cubes.

A few things to consider.

Does the leaf level of your 2 million member dimension contain a lot of
member properties to support your 16 virtual dimensions. If so, I'd
recommend that you consider redesigning this dimension so that you don't
have the leaf level, and instead have a series of regular dimensions that
represent what are currently your virtual dimensions. For example, suppose
this is a customer dimension. Instead of storing 2 million leaf members and
member properties for each member, consider breaking the dimension apart and
building demographic dimensions (gender, age, income, etc). If your users
are really interested in analysis along what are currently your virtual
dimensions, this could keep you from having to have a "customer" level. Of
course, if your users actually want to drill-down to this level of detail on
a dimension, you may find that this not an option for you, but at least it's
something to consider.

Also, consider drill-through or actions. Would drill-through suffice for
this level of detail?

Another option would be the 64-bit edition of Analysis Services. With this
version, you are not limited to 3GB of RAM for dimension members.

Sean


--
Sean Boon
SQL Server BI Product Unit

--
This posting is provided "AS IS" with no warranties, and confers no rights.
Use of included script samples are subject to the terms specified at
http://www.microsoft.com/info/cpyright.htm.





"Dave Ott" <daveo (AT) accessgeneral (DOT) com> wrote