Distributed Computing Using Java: A Comparison of Two Server Designs

摘要

In this paper we explore theoretically and compare experimentally a new server design with a classical concurrent server design implemented in Java. We propose a new concurrent data structure, called parallel hash tables, for synchronizing multithreaded data access. We prove theoretically the upper limit of conflict probabilities in the parallel hash table. To empirically evaluate its benefit, we compare the performance of a TCP multi-threaded parallel hash table-based server to a classical TCP multi-threaded shared buffer-based server. We test them on a network of 35 workstations running Windows NT.