Placing Replicated Data to Reduce Seek Delays

In many environments, seek time is a major component of the disk access time. In this paper we introduce the idea of replicating data on a disk to reduce the average seek time. Our focus is on the problem of placing replicated data. We present several techniques for replica placement and evaluate their performance using trace-driven simulations. 1 Introduction Random access time is a major factor that degrades the performance of disks. As improvements in CPU and memory speeds continue to outpace improvements in disk speeds, the importance of reducing random access times increases [Bitton 87]. Typically, seek time constitutes nearly half of the random access time, the rest being rotational latency and channel contention [Bitton 88]. Average seek time can be reduced significantly by replicating some data on the disk and spreading the replicas across the disk's surface. To satisfy a request for replicated data, a suitable replica, e.g. the one nearest the disk head, can be retrieved..