Types of cubes that can be created by Cognos

[lifzgud]

Hi,

I am using Cognos 8 currently.
I wanted to know about the types of cubes that can be created in cognos and the various tolls which build them.eg I know that DMR can be done by FM and its ROLAP if i'm correct.
I am looking for a more detailed list like which tool can build what types of cubes and what are their advantages and disadvantages .
Please help me out.Really short of time.

[dlafrance]

You can use Cognos Transformer to produce Powercubes

You can also use SQL Server Analysis Services to do cube (version 2005 or 2008 if i can remember well)

We use Transformer and it's good. We use it with FM as a datasource (no DMR in FM)

[lifzgud]

Thanks for your reply.
So I understand Transformer is one tool.What types of cubes can be built from that?(ROLAP,MOLAP,HOLAP etc)
Also please share any other tools that we can use for building cubes.I am looking for an exhaustive list.

[dlafrance]

Hi,

Transformer is doing MOLAP.

As for a list of tools, i will refer you to wikipedia : http://en.wikipedia.org/wiki/MOLAP#Products

[MFGF]

There are two main OLAP cube types you can build with Cognos tools:

1. Powercubes, which are MOLAP, stored as files in the filesystem.
2. TM1 cubes, which again are MOLAP, but are built and held in-memory.

DMR is not really an OLAP cube - it simply allows OLAP-type access to underlying relational data.  Think of DMR as a virtual OLAP environment which translates OLAP queries into SQL.

Regards,

MF.

[tupac_rd]

I am hearing TM1 cubes are the way to go instead of powerplay, is that true... does anybody have experience with TM1...

[peritas-chris]

Lots of experience with TM1 and Express, definitely different technology from Powercubes.  A very complex rules language allows complete control over the cube, but adds development time.  Things like relative time are not as easy as Transformer makes it, among other differences.

The biggest technology difference (in my opinion of course) is the x64 in-memory technology rather than a file such as an mdc.  The way TM1 deals with sparsity (the handling of empty cells) makes a world of difference with cube size and performance.

[ssp]

Think of DMRs as ROLAP.
Think of BOBJ info. cubes as ROLAP.

TM1, Powerplay are MOLAP. However, TM1 handles sparsity, but Powerplay doesn't. If you have zero value cells and a calculation such as "Price * UNIT" is applied where Units = 0, then Powerplay will still run and store the resultant "0" value in a cell. TM1 handles this in the rules engine by writing feeders.
This issue is common. Cognos EP struggled with sparsity, hence you had to build small cubes and run d-links between cubes. You also had to build access tables to suppress 0 valued cells but that didn't make sense if you wanted them exposed to write-back into.

Hope that helps.

[MFGF]

DMR "cubes" are not really ROLAP, being honest. Let's call them OOR (Olap Over Relational) :) They have no aggregate awareness or real intelligence. The new Dynamic Cubes in 10.2 are probably a closer fit to this.

Cheers!

MF.