A Hybrid-based Multi-criteria Approach, for Efficient Workflow Tasks Assignment in Cloud Computing Environment

Cloud computing has emerged as a distribution hub, that offers ubiquitous access, to a shared pool of cloud resources for modeling, executing, and performing big data analyses of businesses as well as scientific workflow applications. The workflow scheduling process is to allocate resources for users’ tasks in a way that satisfies the constraints while optimizing some objectives. One of the most challenging problems in workflow scheduling is determining an optimal number of VMs configurations that can schedule streaming tasks to speed up the execution time while incurring a relatively low cost and energy consumption. Several methodologies have been proposed with the aim of generating good schedules to improve workflow execution. However, most of these existing methodologies on workflow scheduling are improving on relatively simple objectives (cost and makespan) without looking at the bigger problem (VM provisioning), which is mostly the prime causes of high execution cost, makespan, and energy consumption. In this paper, we proposed a novel hybrid-based multi-criteria deadline-constrained scheduler, known as MOW-PSO. MOW-PSO is a fusion of PSO and MOWOS, aiming to determine an optimal number of VMs configuration for a proper task to VM mapping, that finds a suitable solution to the three important yet, conflicting scheduling objectives; energy consumption, execution makespan, and cost without violating user-defined deadlines and budget constraints. Our approach is evaluated using five different representative workflows with four different workload patterns through WorkflowSim. The results prove the effectiveness of our proposed approach over the state-of-the-art algorithms.