Memory-Aware Scheduling in Constrained Environments

  • Unique Paper ID: 182986
  • Volume: 12
  • Issue: no
  • PageNo: 66-80
  • Abstract:
  • This paper addresses the challenge of scheduling workflows in memory-constrained environments where tasks may exceed the available memory on processors. I propose three memory-aware HEFT-based heuristics: HEFTM-BL, HEFTM- BLC, and HEFTM-MM, designed to produce valid schedules by accounting for memory limitations. Experimental evaluations on default and memory-constrained clusters demonstrate the superiority of memory-aware heuristics over the baseline HEFT, which frequently produces invalid schedules. Among the pro- posed methods, HEFTM-MM achieves a 100% success rate for large workflows under severe memory constraints but at the cost of increased makespan and runtime. Additionally, I explore dynamic scenarios with evolving task parameters, highlighting the necessity of adaptive scheduling to prevent execution failures. The study provides a foundation for memory-efficient scheduling in static and dynamic environments and opens avenues for extending the model to handle heterogeneous bandwidths, task variability, and platform dynamics.

Cite This Article

  • ISSN: 2349-6002
  • Volume: 12
  • Issue: no
  • PageNo: 66-80

Memory-Aware Scheduling in Constrained Environments

Related Articles