Hi,
My fact table contains 1 year's data. But the first 6 months needs one
set of dimensions and the next 6 monthts need a different set of
dimensions. So, if I build a cube containing the 1 year's data and all
the dimensions, can I build two virual cubes, one with the first 6
months and the other with last 6 months of data ? How ?

I have not been able to figure out how to restrict data. Can we do
this at all (like the view of a relational table ?)

Thanks a lot.

This can be accomplished by defining dimension security on top of the
virtual cube(s).

public @ the domain below
www.tomchester.net

"mangaraju venuturupalli" <mangaraju (AT) yahoo (DOT) com> wrote

This brings up an excellent question - I know virtual cubes can be used to allow the particular cubes to contain a subset of the measures, but can it be used to allow different cubes to contain a subset of dimensions?

i.e., Measures M1, M2, M3, M4 and M5. Dimensions D1, D2, D2, D3, D4, and D5.

I know its easy to have 2 cubes

Cube1: M1, M2 + D1, D2, D3, D4, and D5
Cube2: M3, M4, M5 + D1, D2, D3, D4, and D5

Then, if you bring them together its the same as if you just had everything in one cube.

Now - what I want to do is have:

Cube1: M1, M2 + D1, D2, D3
Cube2: M3, M4, M5 + D3, D4, and D5.

I've found the statement "Virtual cubes must have one dimension in common". So, what if we add a primary key to the fact table, and use that as a 'fake dimension' to link the two? The fake dimension would be a single dimension with the numbers 1 to a million or however many rows are in the fact table. So, then we would have:

Cube1: M1, M2 + DPK, D1, D2, D3
Cube2: M3, M4, M5 + DPK, D3, D4, D5

Thus each 'event' (like a sale at a store) would have a primary key and the dimensions (salesperson, storeid, etc.) and the facts (price, totalprice, etc.) which would all be linked through the primary key which identifies this as a 'single event'.

What do you think?

It's not true that VCs must have a dim in common, fyi.

public @ the domain below
www.tomchester.net


"OlapGuy" <anonymous (AT) discussions (DOT) microsoft.com> wrote

be used to allow different cubes to contain a subset of dimensions?
as a 'fake dimension' to link the two? The fake dimension would be a single
dimension with the numbers 1 to a million or however many rows are in the
fact table. So, then we would have:
totalprice, etc.) which would all be linked through the primary key which
identifies this as a 'single event'.