hi I have a location dimension that is used by numerous cubes containing all
of the locations in our Company. One of the cubes only contains about half
of the locations in the fact table. In SQL 2000 analytics, the join from the
dimension table to the fact table produced null rows for the locations that
have no match in the fact table, so those rows were easily filtered out.
However in SQL 2005, these rows are included as zeros instead of null, so
"filter nulls" no longer removes these rows. What options are available to
either filter out the "zero" rows, or specify that dimension members with no
matching fact row should be eliminated (inner join).
thanks, Luke

Without looking at your cube definition, I can only guess at a few things:

1) make sure your fact table actually contains NULLs or doesn't write rows
at all for the locations that aren't in the cube.
2) make sure your queries specify NON EMPTY for rows and columns
3) check your formats for the measures that are coming back as 'zero'. I
seem to recall having a similar problem with formats that included trailing
zeroes.
4) 2005 also provides "Non-Empty Behavior" settings for the measures. Try
setting those as well.
5) Calculated fields that don't return NULL when a condition is not met can
wreak havoc.

-T

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