What is difference between MDDB Database cubes and Reporting Cubes

[RTwo]

Need some clarity about MDDB servers.

Why do we need a MDDB server like Oracle Express, MS Analysis Server, Hyperion Essbase, et all. I can understand the advatages of basic things like aggregation, pre summarised data, faster access, cube related security features, building and storing multiple cubes, etc but there is also a cost involved in terms of pushing the data physically to a MDDB server (ETL,/code).

But we can get the same features through a reporting cube, which also gives MOLAP or ROLAP, perhaps at a much less cost!! And if scalability of cubes is the problem, we can also build multiple reporting cubes. Also, one more thought, even if we have a MDDB server, we may still need these OLAP tools and then again build a cube in them. Isn't this sort of duplication and extra work??

 

[Guest_imported]

There seems to be a lot of confusion in the industry about types of databases AND different OLAP models. I was no exception to this. But over some period got these doubts rested. Types of databases are TYPES OF DATABASES and olap models are nothing but architectures.

Types of databases (database engines - MDDB(MOLAP), ROLAP, RDBMS.

Types of OLAPS is more of an architectural issue. Various architectures are – MOLAPs/ROLAPs, HOLAPs, DOLAPs. These are more towards where the cube is or where the “cube-like” processing happen and where the data is.

The advantages of MDDB's you mentioned are strong enough to be the sole drivers to use and MDDB cube. There is a huge performance benefit in using MDDBs. But don't forget that this is true only if the size of the cube is reaosnable. If it starts growing, then you start loosing on the performance front as well.

Having said that, I haven't heard of anything called as "reporting cube" as a TYPE of database. You are probably calling it that way because you are gonna use it for reporting.

MOLAP is nothing but MDDB – cube served by a MDDB engine. Data Source is in cube format, served by a OLAP database engine as a cube.

ROLAP is RELATIONAL database. On the client side end of the database engine, the db engine understands only OLAP queries and not SQL queries. These OLAP queries are then converted into SQL queries and a cube is generated on the fly(only including the data needed for the report). In other words, ROLAP is virtual cube created from a relational database

DOLAP is a desktop based OLAP cube wherein the data retrieved by the OLAP reporting tool is cached into a local cube managed and served by a component of the OLAP reporting tool. In fact many tools either save this data in a flat file dump which is easy to read. It’s nothing but a totally denormalized, merged dump of data from all the dimensional and the fact table.

Any takers?

 

[RTwo]

Appreciate your response. What I meant by 'reporting cubes' were the cubes created in Cognos and Business Objects. They don't use a MDDB server but still make a ROLAP (Business Objects) or MOLAP (Cognos PowerPlay) cube.

I mean they could connect to any relational Database and still form a cube on their own.