Hi,
In figure 2 how would you get data from the cube as of a particular date.
Example 1: I want to know the current count of customer from the fact table which should be 8 but its going to return 20 since there are 20 records.
Example 2: Number of customer as of 2007 which only returns 4 since only 4 records were added to the fact table in 2007 as those were the only customer who had updates.

Thanks.

Vamsi8/25/2008 5:06:38 PM

Although I have played with the monster dimension you discuss, I have settled on a couple of different approaches instead. If the things that describe the dimension item are finite, like a billing plan that is changeable, I may settle on a wide fact tables containing many dimension keys to describe a fewer number of facts. To me, splitting out descriptors by subject into new dimensions such as "salary", "geo", or "education" makes it easier for analysts to interpret at query time and supports a natural migration into an OLAP environment. Alternately, if it is a major data load, like from a market research firm, I will treat the combination of descriptors as one or more fact-less fact tables and use a conformed dimension to join them to an additive fact at query time. This is not dissimilar from what you propose except that it is going through the dimension rather than behaving as a peer. I even call it Fact.Demographics.

Vince3/13/2008 5:29:25 PM