Integrated Prefetching and Caching with Read and Write Requests
All previous work on integrated prefetching/caching assumes that memory reference strings consist of read requests only. In this paper we present the first study of integrated prefetching/caching with both read and write requests. For single disk systems we analyze popular algorithms such as Conservative and Aggressive and give tight bounds on their approximation ratios. We also develop a new algorithm that performs better than Conservative and Aggressive. For parallel disk systems we present a general technique to construct feasible schedules. The technique achieves a load balancing among the disks. Finally we show that it is NP-complete to decide if an input can be served with f fetch and w write operations, even in the single disk setting.
KeywordsElapse Time Optimal Schedule Approximation Ratio Memory Block Disk Operation
Unable to display preview. Download preview PDF.
- 4.Cao, P., Felten, E.W., Karlin, A.R., Li, K.: A study of integrated prefetching and caching strategies. In: Proc. ACM International Conference on Measurement and Modeling of Computer Systems (SIGMETRICS), pp. 188–196 (1995) Google Scholar
- 8.Kimbrel, T., Cao, P., Felten, E.W., Karlin, A.R., Li, K.: Integrated parallel prefetching and caching. In: Proc. ACM International Conference on Measurement and Modeling of Computer Systems (SIGMETRICS) (1996) Google Scholar
- 9.Kimbrel, T., Tomkins, A., Patterson, R.H., Bershad, B., Cao, P., Felten, E.W., Gibson, G.A., Karlin, A.R., Li, K.: A trace-driven comparison of algorithms for parallel prefetching and caching. In: Proc. of the ACM SIGOPS/USENIX Association Symposium on Operating System Design and Implementation (1996)Google Scholar