Distributed Many-to-Many Mapping Algorithm in the Hypercube Network

Server-client mapping is one of the most frequently encountered problems in nearly all the distributed systems, but is usually solved in ad-hoc manner in favor of quick implementation. To facilitate improved design of distributed applications, this paper addresses the issue of many-to-many node mapping in the hypercube. One can imagine that there exists a set of servers containing the data and a large number of clients requesting the data. The problem addressed in this paper is how to match the servers to clients so that the network loads are minimized. In this paper, we present a distributed many-to-many node mapping algorithm which allows scaling the network to very large size. Through simulation, we show that it has performance advantages in a number of areas. First, it reduces the level of congestion at the bottleneck link. Second, it consumes less network resources in terms of the total number of links used and bandwidth usage.