Coordinated Resource Management in a Replicated Objected Server

Abstract

We propose several new techniques for resource management in a replicated object server. By coordinating cache and disk usage among the replicas, these techniques increase throughput and reduce fetch latency. Cache splitting speeds up fetches by avoiding redundant cache entries, effectively increasing the cache size. Coordinated writing coordinates disk writes to ensure that one replica is always available to service fetches. We investigate the performance of a replicated server using these techniques, and we present simulation results showing that these techniques provide substantial performance improvements across a variety of workloads.