Multicast Flow Aggregation in the Internet Backbone

摘要

Current research efforts focus on aggregating unicast sessions. For multicast sessions, only packet scheduling and classifier states but not RSVP states are aggregated, thus, per-flow RSVP states are still stored at interior routers and that brings about the scalability issue. This paper proposes a scheme to aggregate multicast data flows as well as RSVP states in the Internet backbone to resolve the scalability problem. The proposed scheme can assure that the number of aggregate RSVP states is bounded and will not increase proportionally with the number of flows. Evaluation results also confirm that different flows can be guaranteed different shares of network throughput in spite of the aggregation. Thus, the proposed scheme is helpful to resolve the multicast flow aggregation problem.