Study On The Load Balancing Of The Information Platform For Distance Education Experiment

Experiment Teaching is one of the most difficult problems in the distance education nowadays. Students can't do interactive experiment by networks. The problem is well solved in the virtual experiment system for the distance education by combining the virtual experiment technology and the distance education. It is very important to guarantee the stability of the distance information platform for virtual experiment that is a significant part of the virtual experiment system. Load balancing is a prominent technology to guarantee the stability of the platform. It is used in many fields in our society. This paper mainly represents the load balancing technology in cluster domain. The load balancing can be achieved by applications including the middleware, agent, and so on. The operating system software and the hardware can also implement the load balancing. These methods have their advantages and disadvantages. The load balancing algorithms include both the simple methods such as round robin, weighted round robin, least response time and the comparatively complicated methods such as set partition, adaptive dual-threshold dynamic load balancing. The platform on the server is composed of several computers as a cluster to provide the services. The weighted least connection algorithm is chosen due to its excellent effect and practicality. A scheme to value the weight of the load is proposed after the analysis. The subjective load balancing and objective load balancing, initiatively and passively distribution of the load are discussed. The second connection model of load balancing is delivered in order to lower the load balancing in the control device by making the client communicate with the service provider directly after the first connection. The method enhances the stability of the information platform greatly. The stability of the platform has a great impact on the implementing of the load balancing. The network load and server load impact the stability of the platform. By the analysis of that impact, the conclusion can be made as follows: After one client connection fails because of the heavily-loaded network, the failed connection has a little impact on other connections by implementing multithreads on the server, the client whose connection is failed will not wait long by canceling the block. When the data of the server can not be sent to the client in time because of the heavily-loaded network, the probability of leading to thrash phenomenon on the server can be lowered by reducing the data transmitted synchronously between the client and the server. The network load and server load can be lowered by the continuous transmission on the breakpoint.