What to look for when finding the target for table compression

Hi,

I am designing a star schema for a solution and we want to use the capability of the exadata (oracle 11g)

we have standard architecture where reporting will be done at from MOLAP (fed from the star schema).

I have very little experience on the exadata capability though i understand basic feautures what I am looking for any pointer on where should i start looking at .

I mean as rule of thumb what can be put for compression or partition, any obvious big gotcha that i should avoid .