الفهرس | Only 14 pages are availabe for public view |
Abstract Cloud computing is an Internet-based computing with dynamically scalable resources provided to users as services. These services are delivered to users based on Service-Level Agreements (SLA) between the cloud providers and the users. Cloud computing has several data centers at different geographical locations. Each data center has a number of physical hosts configured to users as Virtual Machines (VMs). On the other hand, cloud users have large scale applications with different requirements. They need high-performance computing environment to process their applications. The problem now is how to map the applications’ tasks onto the available heterogeneous VMs to achieve some Quality of Service (QoS) parameters and meet the SLA. Such problem is known as multi-objective scheduling problem. This thesis tackles the multi-objective scheduling problem and presents metaheuristic algorithms to schedule applications’ tasks onto the available VMs in the cloud environment to achieve some objectives in a reasonable amount of time. Further, the meta-heuristic algorithms are hybridized with some other strategies to improve the system performance in terms of QoS parameters including minimizing makespan, total response time and processing cost, as well as maximizing resource utilization. In this thesis, two new hybrid approaches are developed for solving the multiobjective scheduling problem. The first approach is called “Enhanced Particle Swarm Optimization based Chaotic Strategies (EPSOCHO)” while the second approach is called “Enhanced Binary Artificial Bee Colony based Pareto Front (EBABC-PF)”. Further, other two algorithms are developed for achieving load balancing while scheduling tasks in the cloud environment. The new algorithms called Improved Active Monitoring Load Balancer with Hill Climbing Algorithm (IAMLBHC) and Enhanced Load Balancing based on Hybrid Artificial Bee Colony with Enhanced β-Hill Climbing in Cloud (ELBABCEβHC). The implementation and verification of the proposed algorithms are done using CloudSim simulator, WorkflowSim simulator, or CloudAnalyst simulator. The experimental results clearly demonstrate that the proposed approaches achieve better performance in terms of makespan, response time, processing cost and resources utilization compared to the most recent similar algorithms. |