Agent Technology in Load Balancing for Network Applications

Load Balancing is a very important area of research. Initially, the main trend was concerned with parallel systems, but networks showed a new alternative. Several studies expressed the potential strength of combined processing systems. Distributed computing introduced new problems and a new field of research for load balancing. This field was augmented with the usage of the Internet. Networks, on one hand, offer new potentials for immense combined power, but on the other hand, they induce many new factors in load balancing. Many lines of research propose different techniques for load balancing in distributed systems. Most of them are not capable of adapting to the Internet’s world. The most modern approach, studied only in the last three years, employs the use of software agents. These agents introduce ideas and advantages from many different areas, including automata, artificial intelligence, object-oriented programming, etc. The many new characteristics of agents present novel potential solutions to the problem of load balancing for the Web. This paper presents a taxonomy which is used to survey the proposed models and implementations that use software agents to achieve load balancing, mainly in Web applications.