Simulation and analytical approaches for estimating the performability of a multicast address dynamic allocation mechanism

The specification of the Multicast Address Dynamic Client Allocation Protocol (MADCAP) includes a mechanism to assign multicast addresses to the hosts in the Internet environment. MADCAP servers provide multicast address allocation services. A MADCAP client (that is, a host requesting multicast address allocation services via MADCAP) identifies a suitable MADCAP server and sends appropriate messages to the MADCAP server in order to request a multicast address. The mechanism makes the reuse of multicast addresses available with the MADCAP server, through the application of the lease-time concept. Estimating the performance measures of such a system, accurately and reliably, is very important for efficient use of resources and for achieving the QoS (Quality of Service) requirements.This paper proposes two approaches to estimate performability measures of the MADCAP dynamic allocation mechanism. They are essentially, an accurate one by simulation and a fast but approximate one by analytical solution. While the accuracy of the simulation method is limited by the length of the simulation, the analytical method is an approximate one but fast in computation. The detailed simulation model and the analytical method of the MADCAP dynamic allocation mechanism are presented, and the results are compared. Fairly good agreement is established between the simulation and the analytical results, giving rise to accurate and dependable computation of the performability measures.A solution approach using the QBD   model for the performability analysis of the MADCAP mechanism is proposed, which has the computational complexity of the order O(c)O(c). Besides possessing the capability of tackling the problem with large c (the number of available IP multicast addresses is of the order of tens of thousands in the MADCAP server) in a very short time, it also gives numerically stable solution. The impact of the lease-time parameter on the performability of the MADCAP dynamic allocation mechanism is also investigated through a numerical study, and illustrated by graphs.