This Article 
 Bibliographic References 
 Add to: 
Prefetching in File Systems for MIMD Multiprocessors
April 1990 (vol. 1 no. 2)
pp. 218-230

The question of whether prefetching blocks on the file into the block cache can effectively reduce overall execution time of a parallel computation, even under favorable assumptions, is considered. Experiments have been conducted with an interleaved filesystem testbed on the Butterfly Plus multiprocessor. Results of these experiments suggest that (1) the hit ratio, the accepted measure in traditional caching studies, may not be an adequate measure of performance when the workload consists of parallelcomputations and parallel file access patterns, (2) caching with prefetching can significantly improve the hit ratio and the average time to perform an I/O (input/output) operation, and (3) an improvement in overall execution time has been observed in mostcases. In spite of these gains, prefetching sometimes results in increased execution times (a negative result, given the optimistic nature of the study). The authors explore why it is not trivial to translate savings on individual I/O requests into consistently better overall performance and identify the key problems that need to be addressed in order to improve the potential of prefetching techniques in the environment.

[1] R. K. Asbury and D. S. Scott, "Fortran I/O on the iPSC/2: Is there read after write?" inFourth Conf. Hypercube Concurrent Comput. Appl., 1989.
[2] J.-L. Baer and G. R. Sager, "Dynamic improvement of locality in virtual memory systems,"IEEE Trans. Software Eng., vol. SE-2, pp. 54-62, Mar. 1976.
[3] BBN Advanced Computers, "The Butterfly RAMFile system," Tech. Rep. 6351, BBN Advanced Computers, Sept. 1986.
[4] BBN Advanced Computers,Butterfly Products Overview, 1987.
[5] B. T. Bennett and C. May, "Improving performance of buffered DASD to which some references are sequential,"IBM Tech. Disclosure Bull., vol. 24, pp. 1559-1562, no. 3, Aug. 1981.
[6] S. Berbec, A. Shibamiya, S. Togasaki, and H. Yoshida, "Use of direct access storage devices by MVS customers-Guide survey results," inProc. Guide 47 Conf., Nov. 1978, pp. 1121-1138.
[7] BBN Advanced Computers, Inc.,Inside the Butterfly Plus, Cambridge, MA, Oct. 1987.
[8] H. Boral and D. Dewitt, "Database machines: An idea whose time has passed?" inProc 1983 Int. Workshop Database Machines, 1983, pp. 166-187.
[9] T. W. Crockett, "File concepts for parallel I/O," Tech. Rep., ICASE, NASA-Langley, 1988.
[10] D. J. Dewitt, R. H. Gerber, G. Graefe, M. L. Heytens, K. B. Kumar, and M. Muralikrishna, "GAMMA: A high performance dataflow database machine," Tech. Rep. TR-635, Dep. Comput. Sci., Univ. Wisconsin-Madison, Mar. 1986.
[11] D. DeWitt, S. Ghandeharizadeh, and D. Schneider, "A performance analysis of the gamma database machine," inProc. ACM-SIGMOD Int. Conf. Management Data, Chicago, IL, May 1988.
[12] P. Dibble, M. Scott, and C. Ellis, "Bridge: A high-performance file system for parallel processors," inProc. Eighth Int. Conf. Distributed Comput. Syst., June 1988, pp. 154-161.
[13] R. Floyd, "Short-term file reference patterns in a UNIX environment," Tech. Rep. 177, Dep. Comput. Sci., Univ. Rochester, Mar. 1986.
[14] R. Floyd and C. Ellis, "Directory reference patterns in hierarchical file systems,"IEEE Trans. Knowledge Data Eng., vol. 1, June 1989.
[15] R. J. Flynn and H. Hadimioglu, "A distributed hypercube file system," inProc. Third Conf. Hypercube Concurrent Comput. Appl., 1988, pp. 1375-1381.
[16] H. Garcia-Molina and K. Salem, "The impact of disk striping on reliability,"IEEE Database Eng. Bull., vol. 11, pp. 26-39, Mar. 1988.
[17] I. Hu, "Measuring file access patterns in UNIX,"ACM SIGMETRICS Perform. Eval. Rev., vol. 14, pp. 15-20, Aug. 1986.
[18] "Intel beefs up its iPSC/2 supercomputer's I/O and memory capabilities," Electron., Nov. 1988.
[19] iPSC/2 I/O facilities, Intel. Corp., Order no. 280120-001, 1988.
[20] Concurrent I/O application examples, Intel Corp. Background Inform., 1989.
[21] R. H. Katz, J. K. Ousterhout, D. A. Patterson, and M. R. Stonebraker, "A project on high performance I/O subsystems,"IEEE Database Eng. Bull., vol. 11, pp. 40-47, Mar. 1988.
[22] M. Y. Kim, "Synchronized disk interleaving," inProc. IEEE Trans. Comput., vol. C-35, no. 11, pp. 978-988, Nov. 1986.
[23] D. Kotz, "RAPID-Transit: Prefetching and buffering techniques for parallel I/O systems," July 1988, revised Feb. 1989.
[24] R. L. Lee, P. C. Yew, and D. H. Lawrie, "Data prefetching in shared memory multiprocessors," inProc. 1987 Int. Conf. Parallel Process., 1987, pp. 28-31.
[25] M. Livny, S. Khoshafian, and H. Boral, "Multi-disk management algorithms," inProc. SIGMETRICS, pp. 69-77, May 1987.
[26] Thinking machines, Connection Mach. Model CM-2 Tech. Summ., Apr. 1987.
[27] T. Manuel, "Breaking the data-rate logjam with arrays of small disk drives,"Electron., vol. 62, pp. 97-100, Feb. 1989.
[28] J. Menon and M. Hartung, "The IBM 3990 disk cache," inProc. IEEE Compcon, Spring 1988, pp. 146-151.
[29] J. Ousterhoutet al., "Trace-driven analysis of the Unix 4.2 BSD file system," inProc. 10th ACM Symp. on Operating System Principles, Dec. 1985.
[30] J. Ousterhout and F. Douglis, "Beating the I/O bottleneck: A case for log-structured file systems,"ACM Operating Syst. Rev., vol. 23, pp. 11-28, Jan. 1989.
[31] Y. Pang and S. M. Kuck, "Buffering and disk accesses," Tech. Rep. UIUCDCS-R-88-1432, Dep. Comput. Sci., Univ. Illinois at Urbana-Champaign, May 1988.
[32] A. Park and K. Balasubramanian, "Providing fault tolerance in parallel secondary storage systems," Tech. Rep. CS-TR-057-86, Nov. 1986.
[33] D. A. Patterson, G. Gibson, and R. H. Katz, "A case for redundant arrays of inexpensive disks (RAID)," inProc. ACM SIGMOD, Chicago, IL, June 1-3, 1988, pp. 109-116.
[34] P. Pierce, "A concurrent file system for a highly parallel mass storage system, " inFourth Conf. Hypercube Comput. Appl., 1989.
[35] M. L. Powell, "The DEMOS file system," inProc. Sixth Symp. Oper. Syst. Principles, Purdue Univ., Nov. 1977, pp. 33-42.
[36] N. Ragaz and J. Rodriguez-Rosell, "Empirical studies of storage management in a data base system," Tech. Rep. RJ-1834 (26703), IBM San Jose, Oct. 1976.
[37] A. Reddy and P. Banerjee, "An evaluation of multiple-disk I/O systems," inProc. 1989 Int. Conf. Parallel Process., 1989.
[38] A. L. Reddy, P. Banerjee, and S. G. Abraham, "I/O embedding in hypercubes," inProc. 1988 Int. Conf. Parallel Process., vol. 1, 1988, pp. 331-338.
[39] K. Salem and H. Garcia-Molina, "Disk striping," inProc. IEEE Data Eng., pp. 336-342, Feb. 1986.
[40] A. J. Smith, "Sequential program prefetching in memory heirarchies,"IEEE Comput., pp. 7-21, Dec. 1978.
[41] A. J. Smith, "Sequentiality and prefetching in database systems,"ACM Trans. Database Syst., vol. 3, no. 3, pp. 223-247, Sept. 1978.
[42] A. J. Smith, "Input/output optimization and disk architectures: A survey,"Perform. Eval., vol. 1, pp. 104-117, 1981.
[43] A. Smith, "Cache Memories,"Computing Surveys, Vol. 14, No. 3, Sept. 1982, pp. 473- 530.
[44] A. J. Smith, "Disk cache-miss ratio analysis and design considerations,"ACM Trans. Comput. Syst., vol. 3, no. 3, pp. 161-203, Aug. 1985.
[45] M. Stonebraker, "Operating System Support for Database Management,"Comm. ACM, Vol. 24, No. 7, July 1981, pp. 412- 418.
[46] D. Towsley, K. M. Chandy, and J. C. Browne, "Mbdels for parallel processing within programs: Application to CPU:I/O and I/O:I/O overlap,"Commun. ACM, vol. 21, pp. 821-831, Oct. 1978.
[47] D. F. Towsley, "The effects of CPU: I/O overlap in computer system configurations, " inProc. Fifth Annu. Symp. Comput. Architecture, Apr. 1978, pp. 238-241.
[48] K. S. Trivedi, "On the paging performance of array algorithms,"IEEE Trans. Comput., vol. C-26, pp. 938-947, Oct. 1977.
[49] K. S. Trivedi, "Prepaging and applications to array algorithms,"IEEE Trans. Comput., vol. C-25, pp. 915-921, Sept. 1976.
[50] A. Witkowski, K. Chandrakumar, and G. Macchio, "Concurrent IO system for the hypercube multiprocessor,"Proc. 3rd Conf. on Hypercube Concurrent Comput. and Applics., SIAM, Jan. 1988.

Index Terms:
Index Termsconcurrent OS; disc caching; file systems; MIMD multiprocessors; execution time; parallel computation; interleaved file system; Butterfly Plus multiprocessor; hit ratio; performance; caching; prefetching; I/O requests; buffer storage; file organisation; multiprocessing systems; performance evaluation
D.F. Kotz, C.S. Ellis, "Prefetching in File Systems for MIMD Multiprocessors," IEEE Transactions on Parallel and Distributed Systems, vol. 1, no. 2, pp. 218-230, April 1990, doi:10.1109/71.80133
Usage of this product signifies your acceptance of the Terms of Use.