Title :
A distributed heterogeneous supercomputing management system
Author :
Ghafoor, Arif ; Yang, Jaehyung
Author_Institution :
Purdue Univ., West Lafayette, IN, USA
fDate :
6/1/1993 12:00:00 AM
Abstract :
A general management framework for distributed heterogenous supercomputing systems (DHSSs) that is based on an application-characterization technique is presented. The technique uses code profiling and analytical benchmarking of supercomputers. An optimal scheduling of tasks in these systems is an NP-complete problem. The use of network caching to reduce the complexity associated with the scheduling decisions is discussed. An experimental prototype of a DHSS management system is described.<>
Keywords :
multiprocessing programs; network operating systems; open systems; scheduling; supervisory programs; DHSS management system; NP-complete problem; analytical benchmarking; application-characterization technique; code profiling; distributed heterogeneous supercomputing management system; general management framework; network caching; optimal scheduling; scheduling decisions; Buildings; Computer architecture; Computer networks; Degradation; High-speed networks; Power engineering and energy; Power engineering computing; Supercomputers; System performance; Testing;
