I faced some queer problem after I processed my cube.
If I didn't set any aggregation storage seeting, I got the right value.
For Example:

Money

All Date 3000

2004 2000

2005 1000

But, if I did any aggragation seeting like MOLAP or HOLAP, I got triple
or double summation aggregation. Besides that, sometimes each node was
right but (All) got double count.
Foe Example: (wrong)

Money

All Date 6000

2004 4000

2005 2000

or

Money

All Date 6000

2004 2000

2005 1000

Is there any body knowing how to deal with this problem?
Did I miss any important point when designing my cube?

Win 2K platform.
Meta data is in Sql Server.
Sql server is Develop version.
Analysis Service has upgraded to SP3 already.

My guess is that it doesn't have anything to do with your agg's but rather
when you reprocessed you did an incremental process.

Try a full reprocess and see what happens.


"donna" <wasidonna (AT) gmail (DOT) com> wrote

Hello Jesse O... ^^

I tried full reprocess, but got the same result.
If I chose design the aggregation storing seeting, then the process
result will go wrong.

I have no idea about it yet.

Check the joins between the fact table and dimension tables. AS joins by
column names as default and sometimes this is not correct. If you have
matches(by names) between more than one column in the dimension table and
the fact table(and this is not correct) you can get wrong results on
aggregated levels.

Regards
/Thomas Ivarsson

"donna" <wasidonna (AT) gmail (DOT) com> wrote

Hi Thomas ^______^

It Solved now.
Just like what you said, it caused by joined key name problem.

My Fact table and Dimension table used differnt column name in each
table.
I draged key columns of dimension table to fact table to build the
relationship.

I removed those dimensions (with unmatched key column) , and It works
fine.
I think I willl go back to modify my table definition to match those
keys.

Thank U so much!!