Qun Li
Abstract
This paper proposes a parallel file system model under NOWs (network of workstations) environment. According to the features of NOWs, the system incorporates the mechanism of distributed shared memory, particularly the mechanism of COMA (cache only memory access). It links the memory of all nodes into a large cache; each node aggressively uses not only the local memory but also the remote memory of other nodes, which expedites the data accesses dramatically. It also accesses disks in parallel to improve I/O performance. Furthermore, in our model, data are shared naturally and conveniently among nodes as opposed to the traditional parallel file systems. The architecture of the parallel file system and its detailed implementation, such as file read and write, data replacement, file data dissipation, parallel file read and write are presented in this paper.
- 1. Peter F. Corbett, Parallel Access to Files in the Vesta File System, Supercomputing'93, pp. 472-481. Google Scholar
Digital Library
- 2. James C. French, et al., Performance Measurement of the Concurrent File System of the Intel iPSC/2 Hypercube, J. of Parallel and Distributed Computing 17(1-2), pp. 115-121. Google ScholarDigital Library
- 3. M. Dahlin et al., Cooperative Caching: Using Remote Client Memory to Improve File System Performance, Proc. of the first Conference on Operating System Design and Implementation, Nov. 1994. Google ScholarDigital Library
- 4. Ramakrishna Karedla et al., Caching Strategies to Improve Disk System Performance, IEEE Computer, March 1994, pp. 38-46. Google ScholarDigital Library
- 5. David Kotz et al., Caching and Writeback Policies in Parallel File System, J. of Parallel and Distributed Computing 17(1-2), pp. 140-145. Google ScholarDigital Library
- 6. J. del Rosario and A. Choudhary, High Performance I/O for Parallel Computers: Problems and Prospects, IEEE Computer, March 1994, pp. 59-68. Google ScholarDigital Library
- 7. Erik Hagersten, et al., DDM-A Cache-Only Memory Architecture, IEEE Computer Sep. 1992, pp. 44-54. Google ScholarDigital Library
- 8. Thomas E. Anderson, Michael D. Dahlin, et al. Serverless Network File Systems, ACM Transactions on Computer Systems, Vol. 14, No. 1, pp. 41-79, Feb. 1996. Google ScholarDigital Library
- 9. T. Anderson, et al., A Case for NOW, IEEE Micro, pp. 54-64, Feb. 1995. Google ScholarDigital Library
- 10. Susan J. Loverso, Marshall Isman, et al. sfs: A Parallel File System for the CM-5, Proceedings of the Summer 1993 USENIX Conference, pp. 292-307, Cincinnati, Oh.Google Scholar
- 11. Steven A. Moyer and V. S. Sunderam, PIOUS: A Scalable Parallel I/O System for Distributed Computing Environments, 1994 Scalable High Performance Computing Conference, pp. 71-78.Google Scholar
- 12. J. Salmon, CUBIX: Programming Hypercubes without Programming Hosts, Hypercube Multiprocessors 1987, M. T. Heath(ed.), SIAM. 1987.Google Scholar
Index Terms
- BFXM: a parallel file system model based on the mechanism of distributed shared memory
Recommendations
Large files, small writes, and pNFS
ICS '06: Proceedings of the 20th annual international conference on SupercomputingWorkload characterization studies highlight the prevalence of small and sequential data requests in scientific applications. Parallel file systems excel at large data transfers but sometimes at the expense of small I/O performance. pNFS is an NFSv4.1 ...
An Adaptive Cache Coherence Protocol Specification for Parallel Input/Output Systems
Abstract--Caching has been intensively used in memory and traditional file systems to improve system performance. However, the use of caching in parallel file systems and I/O libraries has been limited to I/O nodes to avoid cache coherence problems. In ...
A high-performance distributed parallel file system for data-intensive computations
One of the challenges brought by large-scale scientific applications is how to avoid remote storage access by collectively using sufficient local storage resources to hold huge amounts of data generated by the simulation while providing high-performance ...
Comments