11-28-2012 10:12 AM
I am new to SAS BI as well, so you will for sure be getting better answers. My impression on the difference of the two is that, information map does not contain physical data, rather, it has info on meta data, such as data structure and other characteristics, and it can be built in filters and other queries. So it seems to me more like a combination of meta data and view. While cube is just a fancy name of pivot table, it contains summarized physical data with hierarchy structure and multi-dimensions. The purpose of both is similar, to shield end business users from actual raw data and to facilitate data retrieving process.
11-29-2012 10:44 AM
A key benefit from using cubes is that in the usual case they are composed of pre-calculated summarizations of the detailed data, so getting "summaries" or "totals" is much more efficient. For example, if grocery chain "OrionStar" captures every line item at a cash register, there could be billions of these transaction records every day. Assume every store manager wants to know his sales total every day of the month, a very similar SQL "group by" query would be run against this table.
When a cube is created, the data is summarized once, and the results stored as "records" that can be retrieved by the cube users. The tradeoff is that creating the cube is an added expense, amortized by the reduction in cost for every query that can be efficiently met out of it. Therefore, designing the cube to balance meeting many query demands with a reasonable sized cube is key.
And as Haikuo says, in the process of creating the cube the metadata and data are usually transformed into more "business-friendly" terms.