Evaluating delayed write in a multilevel caching file system

Delayed write in a multilevel file system cache hierarchy offers a way to improve performance in diverse client/server scenarios, such as integrating mass store into a distributed file system or providing distributed file system access over low-speed links. Using file system traces and cache simulations, we explore extensions and modifications to the traditional client-caching model employed in such file systems as AFS, Sprite, and DFS. High cache hit rates at an intermediate cache server-a machine logically interposed between clients and servers that provides cached file service to the clients-combined with high client cache hit rates lend practicality to an integrated mass storage file system. In such a system, magnetic tape or optical-based mass storage devices may be used as a first-class data repository, fronted by disk and RAM caches to offer acceptable access times to the large, but slow, mass storage system. Similarly, a high cache hit rate is necessary for users accessing file systems via low-speed links, where a delayed write intermediate caching server can mediate traffic to make better use of available bandwidth. In an example taken from mobile computing, an intermediate server might be used as a docking station at a user's home. This arrangement would be convenient for users of mobile computers who upload large amounts of data generated while operating in disconnected mode. Simulations of delayed write caching strategies are applicable to both the mass storage and low-speed link scenarios.

[1]  Ernst H. Kristiansen,et al.  Trace-driven simulations for a two-level cache design in open bus systems , 1990, ISCA '90.

[2]  Michael L. Kazar,et al.  Synchronization and Caching Issues in the Andrew File System , 1988, USENIX Winter.

[3]  K. Korner,et al.  Intelligent caching for remote file service , 1990, Proceedings.,10th International Conference on Distributed Computing Systems.

[4]  Peter Honeyman,et al.  Integrating mass storage and file systems , 1993, [1993] Proceedings Twelfth IEEE Symposium on Mass Storage systems.

[5]  K OusterhoutJohn,et al.  Caching in the Sprite network file system , 1988 .

[6]  J. Howard Et El,et al.  Scale and performance in a distributed file system , 1988 .

[7]  Sailesh Chutani,et al.  DEcorum File System Architectural Overview , 1990, USENIX Summer.

[8]  David L. Mills,et al.  Network Time Protocol (Version 3) Specification, Implementation , 1992 .

[9]  John H. Howard,et al.  On Overview of the Andrew File System , 1988, USENIX Winter.

[10]  Derek L. Eager,et al.  Disk cache performance for distributed systems , 1990, Proceedings.,10th International Conference on Distributed Computing Systems.

[11]  Matt Blaze,et al.  Dynamic hierarchical caching in large-scale distributed file systems , 1992, [1992] Proceedings of the 12th International Conference on Distributed Computing Systems.

[12]  Michael Dahlin,et al.  A quantitative analysis of cache policies for scalable network file systems , 1994, SIGMETRICS.

[13]  Mahadev Satyanarayanan,et al.  Disconnected Operation in the Coda File System , 1999, Mobidata.

[14]  Matt Blaze.  NFS Tracing By Passive Network Monitoring , 1992 .

[15]  Daniel A. Muntz,et al.  Multilevel caching in distributed file systems , 1995 .

[16]  David A. Patterson,et al.  Computer Architecture: A Quantitative Approach , 1969 .

[17]  Silvano Maffeis.  Cache management algorithms for flexible filesystems , 1993, PERV.

[18]  Randy H. Katz,et al.  An Analysis of File Migration in a UNIX Supercomputing Environment , 1993, USENIX Winter.

[19]  Sean Quinlan,et al.  A cached WORM file system , 1991, Softw. Pract. Exp..