Not sure what you are trying to achieve here. Clustering is essentially for fault tolerance NOT performance (load balancing), so active/active is not going to share your work load.

With regard to your replicating databases to your third server and reporting on that, it is a good plan but not sure how replication would cope with a failover (never tried this before) but you do need to ensure that your databases lend themselves to replication, ie triggers and active content wont replay on replication (that caught me out once).

Hope this is of use, otherwise if you could explain the why and wherefores like Gail asked, will be easier to advice. 🙂