Hi all,

I am new to DW and started as intern in a campaign marketing firm.
They want to analyse their data to find answers to questions like,

1. agent performance. What made him to perform well.(Analyse the agent behaviour)
2. Analyse the customer behaviour. etc

Currently they have a relational database with tables like Activity, transaction, leads, campaigns, customers, etc. Can I create an OLAP cube with this? Or what are the things i need to look into the table to create a cube.

Do I need to create a DW or just OLAP cubes?

Any help is appreciated. Don't know where to start?

Thanks,
Vinoth

Well, I would say that it depends on the scope of the project (and perhaps the tools involved) With Microsoft BI I would create a couple of starschema's first and use these to create OLAP cube structures. With a premium BI tool like Cognos you have a few more options (ROLAP) cause you have a more advanced metadatamodelling layer to work with.

Depending on complexity and amount of data I can even imagine that you define db views that constitute the starschema's which would mean not having to generate ETL. This would not be a big issue if you only need to generate the cubes overnight.

However, the first step in each case is to define your dimensions and facts. I can imagine you somehow will need to create facttables from 'factless facts'/ 'event' tables in your case

Ties Blom
 
 

Thanks Ties,

Currently I have downloaded the cognos Trial version and might also download Microstrategy. My company wants to see which would be better. Although I see lot of videos to create cubes, I don't see creating facts from factless tables. I would appreciate if can you help me creating a basic cube from the events table. I need a jump start.

Thanks

If you are working on SQL server you might consider using a MS BI solution. This is a far more cost-friendly solution if you have a limited project. What ever solution you choose you will need to follow some regular training to get the basics. for Cognos look also here:

www.cognoise.com  

Ties Blom
 
 

Hi, I am also new to data warehousing and facing this same situation now. But I can do ETL using Informatica but not cubes and not very familiar with the other terminology mentioned above. Can you please help me how you proceeded further in this case? I am not sure how to build the star schemas because the relations between the tables is so messy and confusing.

If you are new, I also recommend the Microsoft BI solution. Microsoft Analysis Services.  It's relatively easy to pick up, and there's a forum here on tek-tips if you get stuck.  

The trouble with doing something right the first time is that nobody appreciates how difficult it was - Steven Wright