Updating Load Information And Adjusting Strategies For Load Balancing Based On Mobile Agent

Load Balancing is a key technology of Distributed System. The goal ofload balancing is to improve the utilization of resource of nodes, bymanaging resource and schedule tasks. In a load balancing system, one nodewhich is busy, can transfer its some tasks into another node which is idle atthat time. In this way, the busy node would not be down by heavy load, andthe efficiency of the tasks and the performance of whole system can beimproved.Load information shows the status of nodes, including the utilization ofCPU, Memory and so on. How this information is exact and timely can affectthe result of load balancing. Most traditional load balancing systems collectand update load information by broadcasting or polling. The former methodwill bring forth much added working for network and nodes. The latter onewill decrease the efficiency of load information collecting. The both way arerealized by message-passing, which is unreliable. Most of them are based onclient/server model, which has fixed roles and single manner, so theirflexibility is low.At present, there are some kinds of load balancing system, which havetheir own strategies for balancing. Every strategy has good performance onlyin its suited environment. There is no any universal strategy now. However,adjusting current strategy for balancing is needed to adapt to the unceasingchange of environment, such as type of tasks, performance of nodes, andstatus of networks.Mobile Agent is a novel technology originated from ArtificialIntelligence and Distributed System. Mobile Agent can be executed and moveamong nodes in a network. Mobile Agent can complete some tasks ondestination nodes asynchronously and independently, according to the mindof its origin node. So Mobile Agent is intelligent. If Mobile Agent technologyis applied in load balancing, the additional load of network and nodes will beless, and the delay of network will be overcome. Mobile Agent can improvethe adaptability, flexibility and others because it has intelligence and canexecute teaks asynchronously and automatically. Our goal in this article is toresolve the two problems above, which are collecting and updating loadinformation, and adjusting strategies for load balancing.In this article, we bring forward LBSMA (Load Balancing System basedon Mobile Agent) at first, and then analyze its architecture and functions. Atthe same time, the applications and principles are described. With the MobileAgent technology, LBSMA can improve the adaptability, flexibility,...