A Heuristic Meta Scheduler for Optimal Resource Utilization and Improved QoS in Cloud Computing Environment

This paper presents a novel Meta scheduler algorithm using Particle Swarm Optimization (PSO) for cloud computing environment that focuses on fulfilling deadline requirements of the resource consumers as well as energy conservation requirement of the resource provider contributing towards green IT. PSO is a populationbased heuristic method which can be used to solve NP-hard problems. The nature of jobs is considered to be independent, non pre-emptive, parallel and time critical. In order to execute jobs in a cloud, primarily Virtual Machine (VM) instances are launched in appropriate physical servers available in a data-center. The number of VM instances to be created across different servers to complete the time critical jobs successfully, is identified using PSO by exploiting the idle resources in powered-on servers. The scheduler postpones the power-up/activation of new servers/hosts for launching enqueued VM requests, as long as it is possible to meet the deadline requirements of the user. The Meta Scheduler also incorporates Backfilling Strategy which improves makespan. The results conclude that the proposed novel Meta scheduler gives optimization in terms of number of jobs meeting their deadlines (QoS) and utilization of computing resources, helping both cloud service consumer as well as cloud service provider.