A Pareto-based Optimizer for Workflow Scheduling in Cloud Computing Environment

Abstract

A scheduling algorithm in cloud computing environment is in charge of assigning tasks of a workflow to cloud’s virtual machines (VMs) so that the workflow completion time is minimized. Due to the heterogeneity and dynamicity of VMs and diversity of tasks size, workflow scheduling is confronted with a huge permutation space and is known as an NP-complete problem; therefore, heuristic algorithms are used to reach an optimal scheduling. While the single-objective optimization i.e., minimizing completion time, proposes the workflow scheduling as a NP-complete problem, multi-objective optimization for the scheduling problem is confronted with a more permutation space. In our pre vious work, we considered single-objective optimization (minimizing the workflow completion time) using Particle Swarm Optimization (PSO) algorithm. The current study aims to present a multi -objective optimizer for conflicting objectives using Gray Wolves Optimizer (GWO) where dependencies exist between workflow tasks. We applied our method to Epigenomics (balanced) and Montage (imbalanced) workflows and compared our results with those of the SPEA2 algorithm based on parameters of Attention Quotient, Max Extension, and Remoteness Dispersal.

Keywords

• cloud computing
• task scheduling
• grey wolf optimizer
• multi -objective optimization
• pareto front
• strength pareto evolutionary algorithm2 (spea2)