Best practices for generating new test environments

I'm planning to use QM to generate the test environments that will be tested by my test plan. The platform coverage list is quite extensive and contains OS's, databases, app servers, browsers, etc... so I'm hoping to use the weightings option to control the occurrence of some attributes over the others.

I've been trying to generate the coverage using the pair-wise interaction option but I keep getting the same # of permutations for each attribute.

Any pointers?

Thank you,
Chakib