Processor Virualization and Migration for PVM

This paper presents an algorithm for scheduling parallel applications in large-scale, multi-user, heterogeneous distributed systems. The approach is primarily targeted at systems that harvest idle cycles in general-purpose workstation networks, but is also applicable to clustered computer systems and massively parallel processors. The algorithm handles unequal processor capacities, multiple architecture types and dynamic variations in the number of processes and available processors. Scheduling decisions are driven by the desire to minimize turn around time while maintaining fairness among competing applications. For efficiency, the virtual processors (VPs) of each application are gang scheduled on some subset of the available physical processors. Scheduling, distributed systems, heterogeneity, process migration, concurrent processing, idle cycle stealing.

[1]  Jack Dongarra,et al.  Heterogeneous network computing , 1991 .

[2]  Jonathan Walpole,et al.  Adaptive load migration systems for PVM , 1994, Proceedings of Supercomputing '94.

[3]  B. Clifford Neuman,et al.  Resource management for distributed parallel systems , 1993, [1993] Proceedings The 2nd International Symposium on High Performance Distributed Computing.

[4]  John K. Ousterhout,et al.  Why Aren't Operating Systems Getting Faster As Fast as Hardware? , 1990, USENIX Summer.

[5]  Jack Dongarra,et al.  Integrated Pvm Framework Supports Heterogeneous Network Computing , 1993 .

[6]  Miron Livny,et al.  Condor-a hunter of idle workstations , 1988, [1988] Proceedings. The 8th International Conference on Distributed.

[7]  Jonathan Walpole,et al.  A user-level process package for PVM , 1994, Proceedings of IEEE Scalable High Performance Computing Conference.

[8]  Jack Dongarra,et al.  PVM: Experiences, current status and future direction , 1993, Supercomputing '93. Proceedings.