ABSTRACT

Due to the unprecedented scale and dynamic nature of the Internet, popular Internet services are often run on Web clusters so as to guarantee the scalability and availability of the services. Web clusters rely on load balancing technologies to distribute the service traffic between the cluster servers transparently to clients. In this chapter, we define the load balancing problem on Internet servers and discuss the requirements of service traffic for load balancing strategies. We present a survey of hardware and software approaches in both server and network sides. The concept and principle of load balancing have long been studied in a general context of parallel and distributed systems. We conclude this chapter by presenting a unified W5 model of the load balancing problem for understanding its key issues, in particular, the rationale behind load balancing strategies in different types of the systems.