Multi-Objective Accelerated Particle Swarm Optimization Technique for Scientific workflows in IaaS cloud

Efficient workflow scheduling with multi-objective optimizations for various scientific workflows is a challenging issue in the cloud environment. The cloud providers often try to deploy the tasks to the suitable VM (Virtual Machine) instances while meeting the QoS constraints of the workflow, such as deadline and budget. However, the QoS constraints are conflicted with each other, i.e. the execution speeds of the cheaper VM instances are slower than the expensive VM instances. Furthermore, the existing scheduling strategies minimize one of the objectives of workflow scheduling such as minimizing the total execution time or cost while meeting a QoS constraint. To overcome the above-mentioned problem, in this paper we propose a multi-objective workflow scheduling strategy referred to MAPSO. The algorithm devises an efficient strategy to select the best-fit VM instance for each task based on the accelerated particle swarm optimization technique. This may minimize the total execution time and cost of the workflow while meeting multiple QoS constraints. The algorithm also devises an efficient strategy to find an optimal schedule of the tasks which may maximize the throughput of the servers. We simulate and compare the MAPSO algorithm with the current state-of-arts-algorithms over various scientific workflows.