Improved Algorithms Based On LVS Dynamic Feedback And Balancing Load

In recent years with popular of the global the Internet and the advent of the 3G era, people are living more and more dependent on the network, so the requirements of the quality of network service will become increasingly important. Linux virtual server could organize multiple servers into one large virtual server cluster using load balancing technology in the Linux environment. Its technology is through the LAN or WAN, combine multiple servers which could provide separate services together, which would location the massive customer requests from Internet reasonable to the server with the scheduling algorithm, and return the results to the requesting client as quickly as possible and accurate. Thus, the zero-latency cluster servers and high reliability became its development goals.The papers is for research purposes, Linux virtual services were studied. Including the study of its mechanisms and methods to achieve load balancing and analyzes the ten scheduling algorithm of IPVS software. As for the disadvantages of exist scheduling algorithms that can not dynamically acquire the current state of the server real-time to change the weights, write python scripts Client and server are running on the client and server-side scheduler in the Linux environment. the scripting runs on the server-side is used to collect the information of server to calculate the current server load weights and give the new weights to the scheduler client requests port 9090 by sending the Post. While the scripting runs on the algorithm. The arrival of the next client request will be assigned with the new weights, thereby improving the utilization of the entire cluster and system throughput capacity and fast response. Finally, the design of Web Application Stress Tool is to make a test of the cluster system software emulation, And record the updates records of weight and the load situation. Test results show that, after considering reasonable improvement of weight of the current situation of server load weights, can effectively prevent and reduce access latency tilt load and increase the throughput of the system.