Research And Application Of Load Balancing Technology On Instant Communication Education Platform

Education network is a new educational and training means based on Internet communication for learning and education. Most current commercial network education platforms are based on B/S, C/S models which main drawbacks are worse interactivity, worse real time, and the high pressure on servers. Instant Communication Education Platform provides a good solution to address these issues.Instant Communication Education Platform is an interactive education service platform running on the Internet based on hybrid P2P technology as the expansion and extension of the school in reality. Communication model of platform adopts hybrid P2P structure with the use of a number of high-performance servers as super nodes. When the users login the platform, they can not only get numerous educational resources provided by the servers, but also exchange and study with others in peer-to-peer at real time. As it need not interact directly through servers between users (peers), the server load is reduced. However, as the platform is the structure with more servers, one of which as super node not only exists as an index server, but also takes responsibility of an application server, therefore it ensures that the server load balancing become an issue must be solved.In this paper, load balancing technology is used in Instant Communication Education Platform, providing the load balancing program applicable to the hybrid P2P network structure.This paper deeply studies basic principle of load balancing technology, load balancing methodology based on DNS, NAT, semi-join, encapsulation and IP packet transmit and usual scheduling algorithms including round robin scheduling, weighted round-robin scheduling, least-connection scheduling, weighted least-connection scheduling, destination hashing scheduling and source hashing scheduling. According to the characteristics of Instant Communication Education Platform, this paper makes a load balancing design for the platform which divides the super node into clusters, and increases load balancer to schedule users' logon requests. In the design of scheduling algorithm, a dynamic feedback adaptive scheduling algorithm based on genetic algorithm is provided, which is based on the negative feedback principle, taken into account of all factors impacted on server load on platform, assigns the current users login requests with weighted round-robin, and make dynamic adjustments of the weights regularly on the feedback load information from the server to calculate the weights of each super nodes in the next period with genetic algorithms. The program has carried out on the Instant Communication Education Platform. The results of experiments and applications show the validity of the load balancing design program for the Instant Communication Education Platform, which has been implemented in "Fangda" Instant Communication Education Platform.