ABSTRACT
In heterogeneous and dynamic environments, efficient execution of parallel computations can require mappings of tasks to processors whose performance is both irregular (because of heterogeneity) and time-varying (because of dynamicity). While adaptive domain decomposition techniques have been used to address heterogeneous resource capabilities, temporal variations in those capabilities have seldom been considered. We propose a conservative scheduling policy that uses information about expected future variance in resource capabilities to produce more efficient data mapping decisions. We first present techniques, based on time series predictors that we developed in previous work, for predicting CPU load at some future time point, average CPU load for some future time interval, and variation of CPU load over some future time interval. We then present a family of stochastic scheduling algorithms that exploit such predictions of future availability and variability when making data mapping decisions. Finally, we describe experiments in which we apply our techniques to an astrophysics application. The results of these experiments demonstrate that conservative scheduling can produce execution times that are both significantly faster and less variable than other techniques.
- {1} http://cs.uchicago.edu/~lyang/Load.Google Scholar
- {2} http://trochim.human.cornell.edu/kb/stat_t.htm.Google Scholar
- {3} Allen, G., Benger, W., Dramlitsch, T., Goodale, T., Hege, H.-C., Lanfermann, G., Merzky, A., Radke, T., Seidel, E. and Shalf, J., Cactus Tools for Grid Applications, Cluster Computing, 4 (2001) 179-188. Google ScholarDigital Library
- {4} Allen, G., Benger, W., Goodale, T., Hege, H.-C., Lanfermann, G., Merzky, A., Radke, T., Seidel, E. and Shalf, J., The Cactus Code: A Problem Solving Environment for the Grid. Proceedings of the Ninth IEEE International Symposium on High Performance Distributed Computing (HPDC9), Pittsburgh, 2000. Google ScholarDigital Library
- {5} Allen, G., Goodale, T., Lanfermann, G., Radke, T., Seidel, E., Benger, W., Hege, H.-C., Merzky, A., Masso, J. and Shalf, J., Solving Einstein's Equations on Supercomputers, IEEE Computer, 32 (1999) 52-58. Google ScholarDigital Library
- {6} Arabe, J.N.C., Beguelin, A., Loweamp, B., Seligman, E., Starkey, M. and Stephan, P., Dome: Parallel Programming in a Heterogeneous Multi-user Environment. Carnegie Mellon University, School of Computer Science, 1995.Google ScholarDigital Library
- {7} Berman, F., Wolski, R., Figueira, S., Schopf, J. and Shao, G., Application-Level Scheduling on Distributed Heterogeneous Networks. Supercomputing'96, 1996. Google ScholarDigital Library
- {8} Czajkowski, K., Foster, I., Kesselman, C., Sander, V. and Tuecke, S., SNAP: A Protocol for Negotiating Service Level Agreements and Coordinating Resource management in Distributed Systems. 8th Workshop On Job Scheduling Strategies for Parallel Processing, Edinburgh, Scotland, 2002. Google ScholarDigital Library
- {9} Dail, H.J., A Modular Framework for Adaptive Scheduling in Grid Application Development Environments. Computer Science, University of California, California, San Diego, 2001.Google Scholar
- {10} Dinda, P.A., Online Prediction of the Running Time of Tasks, Cluster Computing, 5 (2002). Google ScholarDigital Library
- {11} Dinda, P.A., A Prediction-based Real-time Scheduling Advisor. Proceedings of the 16th International Parallel and Distributed Processing Symposium (IPDPS 2002), 2002. Google ScholarDigital Library
- {12} Dinda, P.A. and O'Hallaron, D.R., The Statistical Properties of Host Load. The Fourth Workshop on Languages, Compilers, and Run-time Systems for Scalable Computers (LCR 98), Pittsburgh, PA, 1998, pp. 319-334. Google ScholarDigital Library
- {13} Dinda, P.A. and O'Hallaron, D.R., Host Load Prediction Using Linear Models, Cluster Computing, 3 (2000). Google ScholarDigital Library
- {14} Dinda, P.A. and O'Hallaron, D.R., Realistic CPU Workloads Through Host Load Trace Playback. Proc. 5th Workshop on Languages, Compilers, and Run-time Systems for Scalable Computers (LCR 2000), Springer LNCS 1915, Rochester, NY, 2000, pp. 265-280. Google ScholarDigital Library
- {15} Figueira, S.M. and Berman, F., Mapping Parallel Applications to Distributed Heterogeneous Systems. University of Californian, San Diego, 1996.Google Scholar
- {16} Fox, G.C., Johnson, M.A., Lyzenga, G.A., Otto, S.W., Salmon, J.K. and Walker, D.W., Solving Problems on Concurrent Processors, Prentice-Hall, 1988. Google ScholarDigital Library
- {17} Fox, G.C., Williams, R.D. and Messina, P.C., Parallel Computing Works, Morgan Kaufmann, 1994, 977 pp.Google Scholar
- {18} Gehring, J. and Reinefeld, A., Mars: A Framework for Minimizing the Job Execution Time in a Metacomputing Environment, Future Generation Computer Systems, 12(1) (1996) 87-99. Google ScholarDigital Library
- {19} Kumar, S., Das, S.K. and Biswas, R., Graph Partitioning for Parallel Applications in Heterogeneous Grid Environments. submitted to International Parallel and Distributed Processing Symposium (IPDPS 2002), Florida, 2002. Google ScholarDigital Library
- {20} Kumar, S., Maulik, U., Bandyopadhyay, S. and Das, S.K., Efficient Task Mapping on Distributed Heterogeneous System for Mesh Applications. International workshop on Distributed Computing (IWDC 2001), Calcutta, India, 2001.Google Scholar
- {21} Liu, C., Yang, L., Foster, I. and Angulo, D., Design and Evaluation of a Resource Selection Framework for Grid Applications. Proceedings of the 11th IEEE International Symposium on High-Performance Distributed Computing (HPDC 11), Edinburgh, Scotland, 2002. Google ScholarDigital Library
- {22} Ripeanu, M., Iamnitchi, A. and Foster, I., Performance Predictions for a Numerical Relativity Package in Grid Environments, International Journal of High Performance Computing Applications, 15 (2001). Google ScholarDigital Library
- {23} Schopf, J.M., Performance Prediction and Scheduling for Parallel Applications on Multi-User Cluster. Department of Computer Science and Engineering, University of California San Diego, San Diego, 1998, pp. 247.Google Scholar
- {24} Schopf, J.M., A Practical Methodology for Defining Histograms for Predictions and Scheduling. ParCo'99, 1999.Google Scholar
- {25} Schopf, J.M. and Berman, F., Stochastic Scheduling. SuperComputing'99, Portland, Oregon, 1999. Google ScholarDigital Library
- {26} Smith, W., Foster, I. and Taylor, V., Predicting Application Run Times Using Historical Information. Proceedings of the IPPS/SPDP'98 Workshop on Job Scheduling Strategies for Parallel Processing, 1998. Google ScholarDigital Library
- {27} Turgeon, A., Snell, Q. and Clement, M., Application Placement Using Performance Surface. HPDC2000, Pittsburgh, Pennsylvania, 2000. Google ScholarDigital Library
- {28} Vazhkudai, S., Schopf, J.M. and Foster, I., Predicting the Performance of Wide Area Data Transfer. 16th Int'l Parallel and Distributed Processing Symposium (IPDPS 2002), Fort Lauderdale, Florida, 2002. Google ScholarDigital Library
- {29} Weissman, J.B. and Zhao, X., Scheduling Parallel Applications in Distributed Networks, Journal of Cluster Computing, 1 (1998) 109-118. Google ScholarDigital Library
- {30} Wolski, R., Dynamically Forecasting Network Performance Using the Network Weather Service, Journal of Cluster Computing (1998). Google ScholarDigital Library
- {31} Wolski, R., Spring, N. and Hayes, J., The Network Weather Service: A Distributed Resource Performance Forecasting Service for Metacomputing, Journal of Future Generation Computing Systems (1998) 757-768. Google ScholarDigital Library
- {32} Yang, L., Foster, I. and Schopf, J.M., Homeostatic and Tendency-based CPU Load Predictions. International Parallel and Distributed Processing Symposium (IPDPS2003), Nice, France, 2003. Google ScholarDigital Library
- {33} Yang, Y. and Casanova, H., RUMR: Robust Scheduling for Divisible Workloads. Proceedings of the 12th IEEE Symposium on High Performance and Distributed Computing (HPDC-12), Seattle, 2003. Google ScholarDigital Library
- {34} Yang, Y. and Casanova, H., UMR: A Multi-Round Algorithm for Scheduling Divisible Workloads. Proceedings of the International Parallel and Distributed Processing Symposium (IPDPS'03), Nice, France, 2003. Google ScholarDigital Library
Recommendations
Early-release fair scheduling
Euromicro-RTS'00: Proceedings of the 12th Euromicro conference on Real-time systemsWe present a variant of Pfair scheduling, which we call early-release fair (ERfair) scheduling. Like conventional Pfair scheduling, ERfair scheduling algorithms can be applied to optimally schedule periodic tasks on a multiprocessor system in polynomial ...
EDZL Scheduling Analysis
ECRTS '07: Proceedings of the 19th Euromicro Conference on Real-Time SystemsA schedulability test is derived for the global Earliest Deadline Zero Laxity (EDZL) scheduling algorithm on a platform with multiple identical processors. The test is sufficient, but not necessary, to guarantee that a system of independent sporadic ...
Scheduling jobs under decreasing linear deterioration
This paper considers the scheduling problems under decreasing linear deterioration. Deterioration of a job means that its processing time is a function of its execution start time. Optimal algorithms are presented respectively for single machine ...
Comments