Abstract
In this paper we explore the performance of various aspects of gang scheduling designs. We developed an event-driven simulator of a vanilla gang scheduler that relies on the Distributed Hierarchical Control (DHC) structure. We also developed three variations of the vanilla gang scheduler that rely on a push-down heuristic and on two job-migration schemes to decrease response times by reducing processor idle time. We evaluated the gang schedulers on a compiled, one-month long history of jobs from the Cornell Theory Center that was scheduled by EASY-LL, a particular version of LoadLeveler with backfilling. Our results demonstrate the significant performance improvements that can be achieved with gang scheduling. They also show the performance impact of various aspects in the design of gang schedulers. We identify and discuss the potential benefits of several approaches for addressing a number of gang scheduling issues that, under certain workload conditions, become important in practice. Our techniques include heuristics for mapping jobs to processors and for choosing time quanta, block paging for reducing memory overheads, and the allocation of multiple time-slices to smaller jobs per timeplexing cycle.
Preview
Unable to display preview. Download preview PDF.
References
T. Beretvas and W. H. Tetzlaff. Paging enhancements in VM/SP HPO 3.4. Technical Report TB GG22-9467, IBM Washington Syst. Center, May 1984.
D. L. Eager, J. Zahorjan, and E. D. Lazowska. Speedup versus efficiency in parallel systems. IEEE Trans. Comp., 38:408–423, March 1989.
D. G. Feitelson. Packing schemes for gang scheduling. In Job Sched. Strategies for Parallel Processing, D. G. Feitelson and L. Rudolph (eds.), pages 89–110. Springer-Verlag, 1996. LNCS Vol. 1162.
D. G. Feitelson and L. Rudolph. Distributed hierarchical control for parallel processing. Computer, pages 65–77, May 1990.
D. G. Feitelson and L. Rudolph. Mapping and scheduling in a shared parallel environment using distributed hierarchical control. In Proc. International Conf. Parallel Processing, volume I, pages 1–8, August 1990.
D. G. Feitelson and L. Rudolph. Gang scheduling performance benefits for finegrain synchronization. J. Parallel and Distr. Comp., 16(4):306–318, December 1992.
D. G. Feitelson and L. Rudolph. Evaluation of design choices for gang scheduling using distributed hierarchical control. J. Parallel and Distr. Comp., 35:18–34, 1996.
H. Franke, P. Pattnaik, and L. Rudolph. Gang scheduling for highly efficient distributed multiprocessor systems. In Proc. Frontiers'96, 1996.
A. Hori, H. Tezuka, Y. Ishikawa, N. Soda, H. Konaka, and M. Maeda. Implementation of gang-scheduling on workstation cluster. In Job Sched. Strategies for Parallel Processing, D. G. Feitelson and L. Rudolph (eds.), pages 126–139. Springer-Verlag, 1996. LNCS Vol. 1162.
S. G. Hotovy. Workload evolution on the Cornell Theory Center IBM SP2. In Job Sched. Strategies for Parallel Processing, D. G. Feitelson and L. Rudolph (eds.), pages 27–40. Springer-Verlag, 1996. LNCS Vol. 1162.
S. G. Hotovy. Personal communication. 1997.
S. G. Hotovy, D. J. Schneider, and T. O'Donnell. Analysis of the early workload on the Cornell Theory Center IBM SP2. In Proc. ACM SIGMETRICS Conf. Measurement and Modeling of Comp. Syst., pages 272–273, May 1996.
N. Islam, A. Prodromidis, M. S. Squillante, L. L. Fong, and A. S. Gopal. Extensible resource mangement for cluster computing. In Proc. International Conf. Distr. Comp. Syst., May 1997.
N. Islam, A. Prodromidis, M. S. Squillante, A. S. Gopal, and L. L. Fong. Extensible resource scheduling for parallel scientific applications. In Proc. Eighth SIAM Conf. Parallel Processing for Scientific Comp., March 1997.
T. Kimbrel, A. Tomkins, R. H. Patterson, B. Bershad, P. Cao, E. W. Felten, G. A. Gibson, A. R. Karlin, and K. Li. A trace-driven comparison of algorithms for parallel prefetching and caching. In Proc. USENIX Symp. Operating Syst. Design and Implementation (OSDI), pages 19–34, October 1996.
V. M. Lo. Heuristic algorithms for task assignment in distributed systems. IEEE Trans. Comp., 37(11):1384–1397, November 1988.
T. C. Mowry, A. K. Demke, and O. Krieger. Automatic compiler-inserted I/O prefetching for out-of-core applications. In Proc. USENIX Symp. Operating Syst. Design and Implem. (OSDI), pages 3–17, October 1996.
J. K. Ousterhout. Scheduling techniques for concurrent syst., In Proc. Third International Conf. Distr. Comp. Syst., pages 22–30, October 1982.
V. G. Peris, M. S. Squillante, and V. K. Naik. Analysis of the impact of memory in distributed parallel processing systems. In Proc. ACM SIGMETRICS Conf. Measurement and Modeling of Comp. Syst., pages 5–18, May 1994.
J. Skovira, W. Chan, H. Zhou, and D. Lifka. The EASY-LoadLeveler API project. In Job Sched. Strategies for Parallel Processing, D. G. Feitelson and L. Rudolph (eds.), pages 41–47. Springer-Verlag, 1996. LNCS Vol. 1162.
M. S. Squillante, F. Wang, and M. Papaefthymiou. An analysis of gang scheduling for multiprogrammed parallel computing environments. In Proc. Annual ACM Symp. Parallel Algorithms and Architectures (SPAA), pages 89–98, June 1996.
M. S. Squillante, F. Wang, and M. Papaefthymiou. Stochastic analysis of gang scheduling in parallel and distributed syst., Perf. Eval., 27&28:273–296, 1996.
W. H. Tetzlaff. Paging in the VM/XA system product. CMG Trans., 66:55–64, 1989.
W. H. Tetzlaff. Paging in VM/ESA. In Proc. CMG'91 Conf., pages 723–734, 1991.
W. H. Tetzlaff and T. Beretvas. Paging in VM/370 operating systems. CMG Trans., 53:65–76, 1986.
W. H. Tetzlaff, T. Beretvas, W. M. Buco, J. Greenberg, D. R. Patterson, and G. A. Spivak. A page-swapping prototype for VM/HPO. IBM Syst. J., 26:215–230, 1987.
W. H. Tetzlaff and R. Flynn. A comparison of page replacement algorithms. In Proc. CMG'92 Conf., pages 1136–1143, 1992.
W. H. Tetzlaff, M. G. Kienzle, and J. A. Garay. Analysis of block-paging strategies. IBM J. Res. and Devel., 33(1):51–59, January 1989.
F. Wang. Multiprogramming for parallel and distributed systems. PhD thesis, Computer Science Department, Yale University, 1997.
F. Wang, H. Franke, M. Papaefthymiou, P. Pattnaik, L. Rudolph, and M. S. Squillante. A gang scheduling design for multiprogrammed parallel computing environments. In Job Sched. Strategies for Parallel Processing, D. G. Feitelson and L. Rudolph (eds.), pages 111–125. Springer-Verlag, 1996. LNCS Vol. 1162.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1997 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Wang, F., Papaefthymiou, M., Squillante, M. (1997). Performance evaluation of gang scheduling for parallel and distributed multiprogramming. In: Feitelson, D.G., Rudolph, L. (eds) Job Scheduling Strategies for Parallel Processing. JSSPP 1997. Lecture Notes in Computer Science, vol 1291. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-63574-2_26
Download citation
DOI: https://doi.org/10.1007/3-540-63574-2_26
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-63574-1
Online ISBN: 978-3-540-69599-8
eBook Packages: Springer Book Archive