Complexity versus quality: a trade-off for scheduling workflows in heterogeneous computing environments

• Published in: The Journal of supercomputing Vol. 79; no. 1; pp. 924 - 946
• Main Author: Sirisha, D.
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.01.2023; Springer Nature B.V
• Subjects: Algorithms; Compilers; Complexity; Computation; Computer Science; Critical path; Heuristic; Heuristic scheduling; Interpreters; Processor Architectures; Programming Languages; Schedules; Scheduling; Heuristics; Task scheduling; Heterogeneous computing environments; Makespan; Branch and Bound Strategy; Workflow scheduling; List scheduling
• ISSN: 0920-8542; 1573-0484
• DOI: 10.1007/s11227-022-04687-x

Devising an efficient workflow scheduling algorithm is paramount to explore high performance from Heterogeneous Computing Environments. In this paper, two scheduling algorithms are proposed to minimize the schedule length. A novel list-based heuristic scheduling algorithm namely Global Highest degree Task First (GHTF) algorithm is proposed which focuses on increasing the degree of parallelism to reduce the makespan. The proposed heuristic based Branch and Bound strategy namely Critical Path/Earliest Finish Time (CP/EFT) algorithm is devised to schedule workflows to optimality while minimizing the time complexity for evaluating each state in the search space. The GHTF algorithm has shown improvement in the schedules with no additional overhead in computations. Experimental results disclosed that GHTF algorithm generated better schedules by 5–20 percent while CP/EFT algorithm improved the schedules by 10.76–23.45 percent compared to classical list scheduling algorithms. Moreover, CP/EFT algorithm outperformed GHTF algorithm by 5 percent.