Richard is right - because clustering is a 'Shared Nothing' architecture - only one node at a time would be able to host a particular database.

Imagine the extreme scenario where your single database server held a single database - and then you wanted to scale this - clustering would give you no benefits at all.

To take advantage of active/active clustering to scale, you would need to divide up the data into multiple databases and change the application to 'know' where data was - but that's a very different scenario to what I originally faced.