Research And Implementation Of Load Balancing In Data Center

In recent years,with development of the Internet,the number of users of Internet business has increased significantly.The service deployment method change to use a L4 load balancing system to achieve horizontal expansion of services.This also makes the L4 load balancing system an important basic function in modern data centers.The development and maturity of software-defined network technology gives the possibility of further improvement of the L4 load balancing system.This article studies the L4 load balancing system from two aspects:(1)A L4 load balancing system capable of sensing server status.The first is to improve the load distribution part of the L4 load balancing system by collecting background server information and combining machine learning algorithms.And it was deployed in hardware switches and experiments were performed to verify that the method can effectively improve the load distribu-tion effect of the L4 load balancing system.(2)A semi-state L4 load balancing system.The second is to improve the connection state preservation method of the L4 load balancing system by de-signing a connection state compression scheme based on Bloom filter.Experi-ments on hardware switches can show that this solution can effectively increase the number of connections that a single box can maintain,and can effectively reduce the impact of resource exhaustion attacks on the L4 load balancing sys-tem.