Fault Aware Load Balancing and Learning Based Resource Allocation in Cloud

Abstract

Resource allocation and load balancing in a cloud environment is a problem of interest in recent years. With the increase in number of requestover the data centers and size of cloud infrastructure over time, increasing the load unbalancing and power consumption of the data center. So, the requests need to be balanced in such manner having a more effective strategy for resources utilization, request failure, and improved power consumption. Cloud computing made it more complicated with respective to requests types that affect the performance of system. In general, resource allocation and load balancing algorithm chooses an objective function to select a host with least resource utilization, power consumption to optimize the system performance and provide high Quality of Service. The objective of this thesis is to bridge the gap between these research directives, 1) load balancing of requests in faulty cloud environment to improve average resource utilization, reduce request failure count, request delay and unbalancing of load among servers, 2) resource allocation and scheduling of requests in faulty cloud where the fault can be due to hardware of software and to improve the performance of system by reducing failure count, failure probability and improve system reliability, 3) improve power efficiency of system by fault and power aware scheduling algorithm for cloud infrastructure. In chapter three, set of approaches for request load balancing in a faulty cloud environment to effectively utilizes the server capabilities and resources are proposed. We have also studied and analyzed the algorithm over high request load over a system with various test suites and various request types from the dataset. The results obtained with our approach are compared and analyzed with state of art existing algorithm from literature to study the improvement in performance.