Design of client aware scheduler for XEN with enhanced techniques to improve cloud performance

Abstract

Infrastructure as a Service (IaaS) in cloud provides ample scope for various high volume applications to be run on the servers across the WAN, availing a fair service to the end clients. Many effective schedulers have been designed to consider the contention of the computational and communicational resources, which provide a guaranteed effectiveness for resource sharing. However, the vast diversity of client devices in a cloud demand scheduling based on their features and capabilities. Mobile clients, workstations, laptops, PDAs and thin clients in the cloud vary in aspects such as processing power, screen size, battery life, geographical distance and many. Algorithms in the cloud, which are based on client capabilities result in many consumer benefits such as network load balancing, saving battery, reducing latency, efficient processing. In this thesis, the authors propose a client aware credit scheduler for virtualized server setup in a cloud that schedules the client requests based on the client device features and capabilities. Rich Internet Applications (RIA) is proposed, in order for the server to realize the client device capabilities such as the type of client, the battery remaining and the location of the client. Results show that the client-aware credit scheduler is effective in terms of saving energy and reducing response latency.