Optimizing and load balancing for flash crowd to improve quality of service in content delivery network

Abstract

The online access has been increasing rapidly with the digitization of information, cheaper Internet service and affordable devices to access the Internet. This entails for not only handling increasing number of web requests but also meeting Quality of Service (QoS) requirements of end-users. Content Delivery Network (CDN) system is used to make better the performance of origin server by storing the popular contents on surrogate servers. The contents are disseminated to the web users through surrogate servers. The performance of CDN system relies on the selection of appropriate surrogate server to satisfy end-users’ requests. The proposed method named Load Balancing using Neighbors and Utility Computing (LBNUC) takes into account requests arrival rate, load on surrogate servers, end-users’ changing demand and capacity of surrogate servers. The aim is efficient utilization of CDN resources to minimize the time required to serve end-users requests and the cost of servicing requests. This method is also effective in handling of flash crowd situation by monitoring request rate. It handles this situation with support from neighbor surrogate servers and arranging additional resources, if required, through utility computing to meet QoS requirement of end-users.