In this chapter, we first introduced the dimensioning procedure which assigns bandwidth capacity to individual MSNs according to their traffic load. The computation of the traffic load requires three parameters: (1) a network topology; (2) a specific routing algorithm; and (3) an assumed traffic distribution. By iteratively adjusting the bandwidth assignments, the resulting configuration can achieve better routing performance. Based on this configuration method, we evaluated the multicast routing algorithms over a variety of traffic distributions and network topologies. We showed that ICT is the best algorithm for achieving the smallest session rejection rate and BCT is good at producing small diameter trees and is the closest to ICT in achieving high overlay network utilization. Yet it can reject as much as 20% more sessions than ICT in some network configurations. We also quantified the ICT performance with dynamic session configurations and showed that in order to prevent flow disruptions in dynamic sessions, the overlay utilization has to be sacrificed since some nodes that no longer have active local users, have to remain in the tree to support the multicast session. Last, we investigated a hybrid scheme that can reduce computational complexity for large sessions and achieves good overlay utilization.