Research On SSL Server Cluster System Performance Optimization

:SSL (Secure Socket Layer) protocol is used to ensure the security of network communication. So it is widely used in server cluster system to provide a safe data transmission channel between the client and server. Due to the complexity of the SSL protocol authentication scheme and encryption algorithms, the SSL protocol needs taking up a lot of server resources which results in the server performance degradation. To solve the above problem, this thesis mainly studies how to improve server performance in the SSL server cluster. The main contribution can be concluded as follows:(1) Because the cluster structure determines the level of performance, a well-designed server cluster system can greatly improve the overall performance of the cluster. Firstly, this thesis elaborates the SSL server cluster system architecture. Then this thesis elaborates the SSL server cluster overall function. Finally, in order to solve the problem of frequent interruptions of the SSL connection which are caused by the sudden crash of working server in the SSL server cluster, this thesis deep studies the SSL connection recovery mechanism. In this mechanism, the SSL session parameters which have been established are stored in database system, when the server failure occurs, a normal server will extract the necessary session parameters from the database.This mechanism avoids establishing a complete SSL connection, reducing the unnecessary loss of SSL handshake connection, ensuring the stability of the SSL connection, so that the server cluster system can provide more efficient security services.(2) In order to improve the processing power of the cluster system and reduce the response time, this thesis proposes a load balancing algorithm based on service and weights (SW-LBA). Firstly, the dispatcher accepts the requests from clients through the epoll model.Then the servers in the pool collect the load information in real time and send the information to the request dispatchers with dynamic feedback method. In the next stage, the dispatchers calculate the servers’load by combining the servers’ability with servers’real load, and judge the servers state using the RED (random early detection) technique. Finally the dispatcher selects the target server for the static and dynamic request with different ways. The algorithm proposed in this thesis takes the request service type, server processing ability and the server real load condition into consideration, thus it can make full use of the server resources.Theoretical verification and experimental analysis have shown that the method proposed in this thesis has effectively improved servers system performance...