Research On Load Balancing Technology In Parallel Analysis Of IPv6 Network Traffic

With the rapid development of Internet application technology and the constant expansion of the scale, how to protect the network information security has become an important issue need to be resolved in next generation Internet. Parallel analysis based on Network Security System like Firewall System, Intrusion Prevention System and Traffic Detection System has become an effective method to solve network security problems.Among these systems, the first problem to be solved is to ensure load balancing for network data flow. This requires the system to classify packets fast and real-time. At present, for IPv4 networks traffic, the perfect solution has been proposed in domestic and overseas, but for IPv6 networks traffic, no solution has been proposed. In this paper, based on the related technologies, we solve the problem in parallel analysis of IPv6 network traffic as follows:Firstly, considering the static load balancing of networks traffic, a hash algorithm for parallel analysis of IPv6 network traffics is proposed in this paper. This algorithm analyzes the traditional hash algorithm in context, mainly check every byte in IPv6 address to find their changing rule, and ultimately propose a hash algorithm based on bit entropy. By objectively evaluating the algorithm under different hash space and experiments with traditional hash algorithm, superiority and rationality of the algorithm is proved in this paper.Secondly, considering the dynamic load balancing of networks traffic, evaluation criteria of load balancing in parallel analysis is mentioned in this paper, as well as a session-oriented adaptive dynamic load balancing model. Load balancing application object discussed in this paper is an instruction detection system based on multi-core platform. In this system, each processor can be seen as the network data flow detection engine, each detection engine has identical responsibility. The model proposed in this paper, firstly use a static load balancing algorithm to avoid the different packets attributed to the same data flow arriving at different detection engines, then maintain several state table to ensure the model can monitor load information of each detection engines in real-time. Using this model can distribute data flows to a detection engine with lighter load that can achieve the goals for network load balancing.In order to validate the superiority of adaptive load balancing model, we separately add up the modle loaded in system before and after, and use the evaluation criteria of load balancing in parallel analysis which is mentioned in this paper to prove the superiority and rationality of the model.