Which OLAP technology to choose

IBM currently offers 4 different OLAP technologies, all of which offer a specific range of applications :

  • Transformer Power Cube
  • DMR model via Framework Manager
  • TM1
  • Dynamic Cubes

A transformer power cube is a technology that has been around for a long time.  It served it purpose back in the times of Powerplay, but still does today.  Administrators are still able to create transformer cubes, and create a package on them in Framework Manager.  Transformer are very fast, but have a few drawbacks however.  The most important ones are the sizing (2GB is really the maximum, so not too many members is the message), the fact that the cube has to be processed overnight and the missing feature on member attributes.

DMR (Dimensionally modeled relation) is a technology that allows a Framework Manager modeler to create a dimensional model above a relational model created in Framework Manager.  This is especially useful if your data contains too many distinct members (so a power cube would grow too large), but not too much data (as all queries are translated into SQL statements).

TM1 is a product that has been acquired by IBM some time ago already.  It's an in-memory tool that allows you to write back data to the cube.  IBM doesn't really advise doing a lot of reporting on top of a TM1 cube, but it's great to make a planning application for example.  Dit you know that IBM Cognos Insight makes a simple TM1 cube underneath the hood when you drag an excel file into the application?

Dynamic Cubes are the most recent addition to the OLAP repertoire of IBM.  This is also an in-memory technology that focusses on analysis of really large datasets.  I'm talking TB's here.  There are of course some hardware requirements involved here, but it's well worth the investment.  Creating dynamic cubes on small datasets is overkill, for small sets of data you'd be better of creating a power cube or DMR model.

Hope this helps in making your choice.

Cheers!