Enhanced resource provisioning in cloud environment

Abstract

Cloud environments render relatively inexpensive computing resources as a subscription-based service. This service can be effectively utilized by acquiring on-demand computational resources dynamically and the pay-per-use model of the cloud. All the benefits offered by the cloud providers can only be realized by optimal provisioning of the leased resources within the stipulated scheduling time. Using scientific workflows in the cloud environment enables addressing solutions to peta-scale scientific problems. Also its multifold benefits include increasing scalability, dynamic deployment, improved resource utilization and improved performance/cost ratio. Since cloud is not a utopian environment, failures are inevitable that may result in experiencing fluctuations in the delivered performance. Therefore, robust and fault-tolerant scheduling is essential to handle performance variations of cloud resources and failures in the environment. This thesis provides effective fault-tolerant scheduling heuristics for workflows on cloud computing platforms whilst optimizing cost and time. newline