ACM Transactions on Computer Systems (TOCS)
Communications of the ACM
The Thoth System
SOSP '77 Proceedings of the sixth ACM symposium on Operating systems principles
The distributed V kernel and its performance for diskless workstations
SOSP '83 Proceedings of the ninth ACM symposium on Operating systems principles
File servers for network-based distributed systems
ACM Computing Surveys (CSUR)
UIO: a uniform I/O system interface for distributed systems
ACM Transactions on Computer Systems (TOCS)
Performance analysis of file replication schemes in distributed systems
SIGMETRICS '87 Proceedings of the 1987 ACM SIGMETRICS conference on Measurement and modeling of computer systems
Extensions for multi-module records in conventional programming languages
POPL '87 Proceedings of the 14th ACM SIGACT-SIGPLAN symposium on Principles of programming languages
Experimental analysis of layered Ethernet software
ACM '87 Proceedings of the 1987 Fall Joint Computer Conference on Exploring technology: today and tomorrow
Communications of the ACM
Protocols for large data transfers over local networks
SIGCOMM '85 Proceedings of the ninth symposium on Data communications
Design considerations for the transformation of MINIX into a distributed operating system
CSC '88 Proceedings of the 1988 ACM sixteenth annual conference on Computer science
A replicated, distributed file system
EW 2 Proceedings of the 2nd workshop on Making distributed systems work
Unix file access and caching in a multicomputer environment
MSYM'93 Proceedings of the 3rd conference on USENIX MACH III Symposium - Volume 1
Hi-index | 0.02 |
Network file access efficiency is a key issue in a distributed system's performance, especially when many of the network nodes are diskless and rely on a shared network file server. We have designed and implemented a file server that uses the network interprocess communication of the V kernel for file access. This paper describes the basic design of the file server with emphasis on the performance-critical areas. We also give its performance under a variety of workloads and compare these measurements with results predicted by other modeling studies.We conclude that the buffering and disk layout strategies we have used work well under load. Performance results are consistent with a previous modeling study that the file server processor is the most critical resource. However, our experiments with high load were limited by the small amount of buffering on the network interface, i.e. large numbers of packets are dropped at high load giving poorer than predicted performance.