Processor Allocation in Multiprogrammed Distributed-Memory Parallel Computer Systems

In this paper, we examine three general classes of space-sharing scheduling policies under a workload representative of large-scale scientific computing. These policies differ in the way processors are partitioned among the jobs as well as in the way jobs are prioritized for execution on the partitions. We consider new static, adaptive and dynamic policies that differ from previously proposed policies by exploiting user-supplied information about the resource requirements of submitted jobs. We examine the performance characteristics of these policies from both the system and user perspectives. Our results demonstrate that existing static schemes do not perform well under varying workloads, and that the system scheduling policy for such workloads must distinguish between jobs with large differences in execution times. We show that obtaining good performance under adaptive policies requires somea prioriknowledge of the job mix in these systems. We further show that a judiciously parameterized dynamic space-sharing policy can outperform adaptive policies from both the system and user perspectives.

[1]  Honbo Zhou,et al.  The EASY - LoadLeveler API Project , 1996, JSSPP.

[2]  John Zahorjan,et al.  Zahorjan processor allocation policies for message-passing parallel computers , 1994, SIGMETRICS 1994.

[3]  D. Kwak,et al.  LU-SGS implicit algorithm for three-dimensional incompressible Navier-Stokes equations with source term , 1989 .

[4]  Joel H. Saltz,et al.  Data parallel programming in an adaptive environment , 1995, Proceedings of 9th International Parallel Processing Symposium.

[5]  Mark S. Squillante,et al.  Extensible Resource Scheduling for Parallel Scientific Applications , 1997, PPSC.

[6]  Mark S. Squillante,et al.  Performance analysis of job scheduling policies in parallel supercomputing environments , 1993, Supercomputing '93. Proceedings.

[7]  Stephen S. Lavenberg,et al.  Computer Performance Modeling Handbook , 1983, Int. CMG Conference.

[8]  Satish K. Tripathi,et al.  The Processor Working Set and Its Use in Scheduling Multiprocessor Systems , 1991, IEEE Trans. Software Eng..

[9]  Kenneth C. Sevcik.  Characterizations of parallelism in applications and their use in scheduling , 1989, SIGMETRICS '89.

[10]  G. J. A. Stern,et al.  Queueing Systems, Volume 2: Computer Applications , 1976 .

[11]  James Patton Jones.  NAS Requirements Checklist for Job Queuing/Scheduling Software , 1996 .

[12]  Tilak Agerwala,et al.  SP2 System Architecture , 1999, IBM Syst. J..

[13]  T. Pulliam.  Efficient solution methods for the Navier-Stokes equations , 1986 .

[14]  Mark S. Squillante,et al.  Dynamic Partitioning in Different Distributed-Memory Environments , 1996, JSSPP.

[15]  Marios C. Papaefthymiou,et al.  Stochastic Analysis of Gang Scheduling in Parallel and Distributed Systems , 1996, Perform. Evaluation.

[16]  Mark S. Squillante,et al.  Optimal Scheduling of Coarse-Grained Parallel Scientific Applications , 1997, PPSC.

[17]  Mark S. Squillante,et al.  Extensible resource management for cluster computing , 1997, Proceedings of 17th International Conference on Distributed Computing Systems.

[18]  Thomas R. Gross,et al.  Impact of Job Mix on Optimizations for Space Sharing Schedulers , 1996, Proceedings of the 1996 ACM/IEEE Conference on Supercomputing.

[19]  Satish K. Tripathi,et al.  A Comparative Analysis of Static Processor Partitioning Policies for Parallel Computers , 1993, MASCOTS.

[20]  Satish K. Tripathi,et al.  Analysis of Processor Allocation in Multiprogrammed, Distributed-Memory Parallel Processing Systems , 1994, IEEE Trans. Parallel Distributed Syst..

[21]  Bhagirath Narahari,et al.  Scheduling independent tasks on partitionable hypercube multiprocessors , 1993, [1993] Proceedings Seventh International Parallel Processing Symposium.

[22]  Dror G. Feitelson,et al.  Job Characteristics of a Production Parallel Scientivic Workload on the NASA Ames iPSC/860 , 1995, JSSPP.

[23]  Mark S. Squillante,et al.  On the Benefits and Limitations of Dynamic Partitioning in Parallel Computer Systems , 1995, JSSPP.

[24]  David H. Bailey,et al.  The Nas Parallel Benchmarks , 1991, Int. J. High Perform. Comput. Appl..

[25]  Vijay K. Naik,et al.  A Scalable Implementation of the NAS Parallel Benchmark BT on Distributed Memory Systems , 1995, IBM Syst. J..

[26]  Anoop Gupta,et al.  Process control and scheduling issues for multiprogrammed shared-memory multiprocessors , 1989, SOSP '89.

[27]  José E. Moreira,et al.  A Programming Environment for Dynamic Resource Allocation and Data Distribution , 1996, LCPC.

[28]  Marios C. Papaefthymiou,et al.  Performance Evaluation of Gang Scheduling for Parallel and Distributed Multiprogramming , 1997, JSSPP.

[29]  José E. Moreira,et al.  Application-Assisted Dynamic Scheduling on Large-Scal Multi-Computer Systems , 1996, Euro-Par, Vol. II.

[30]  Mark S. Squillante,et al.  Analysis of the Impact of Memory in Distributed Parallel Processing Systems , 1994, SIGMETRICS.

[31]  Sanjeev Setia,et al.  The Interaction between Memory Allocation and Adaptive Partitioning in Message-Passing Multicomputers , 1995, JSSPP.

[32]  ZahorjanJohn,et al.  A dynamic processor allocation policy for multiprogrammed shared-memory multiprocessors , 1993 .

[33]  Giuseppe Serazzi,et al.  Robust Partitioning Policies of Multiprocessor Systems , 1994, Perform. Evaluation.

[34]  Mary K. Vernon,et al.  Use of application characteristics and limited preemption for run-to-completion parallel processor scheduling policies , 1994, SIGMETRICS.

[35]  Laxmi N. Bhuyan,et al.  Parallel algorithms for hypercube allocation , 1993, [1993] Proceedings Seventh International Parallel Processing Symposium.

[36]  Raj Vaswani,et al.  A dynamic processor allocation policy for multiprogrammed shared-memory multiprocessors , 1993, TOCS.