Research On Application Of LVS Based Load Balancing Algorithm In DRC Cluster

With the rapid development of computer technology in recent years, the popularity of mobile Internet reached an unprecedented level, the growth rate of application is more than any time in the past, at the same time, data accumulation also reached a new level. It is undoubted that big data era is the most significant label in today's world.But how to manage those massive, heterogeneous data effectively,and how to tap the data's value fully, and how to achieve the effective integration of different data sources to avoid the "information island".Those problems have become the challenge of existing technical architecture. In 2012, the data-oriented architecture(DOA) has been borned, which is based on "data-oriented and data is as core".The DOA has found a new way to solve the problems of traditional software design in dealing with massive data and "information island". As the core module of DOA, DRC(data register center) bear the work of all data registration, data management and data operation. DRC's access performance has a great influence on DOA, so it is particularly urgent to improve the access performance of the DRC.DRC will be massive, heterogeneous, multi-source data unified definition and registration, forming a huge pool of data resources for a variety of fragmentation and personalized applications to provide data services. In order to improve the access performance of DRC, DRC needs to use cluster technology in the deployment, and the optimization of the weight scheduling algorithm is used to allocate the request. In this paper, by analyzing the characteristics of DRC and the characteristics of LVS scheduling algorithm, the paper proposes an improved LVS DRC cluster scheduling algorithm to optimize the load, in order to improve the data access efficiency of DRC and the throughput of the entire DOA system.The main research contents of this paper are as follows:(1) Research and summarize the development of DOA technology, the development of load balancing technology, and the research of the paper introduces the DRC cluster server access performance related issues, as well as the LVS scheduling algorithm DOA.(2) Analysis of the characteristics of IPVS existing various scheduling algorithms, and aiming at the shortage in the DRC cluster WLC proposed based on SVM improved weight calculation, dynamic adjustment of DRC server pool the server's weight, load balancers scheduling according to the new weight, achieve rational allocation of user requests, enhance the performance of DRC server.(3) To build a small DRC cluster, and the improved algorithm deployed in this paper to the system, the use of pressure testing tools to test the performance of DRC before and after the contrast improvement to make a comparative analysis.The innovations and achievements of this paper are as follows:(1) Based on the DOA architecture, the characteristics of the DRC cluster server's access request structure is studied.Based on the DOA architecture, the characteristics of the core modules of DRC are analyzed, and how to effectively utilize the cluster technology to improve the access performance of DRC system is analyzed.(2) The improved LVS scheduling algorithm based on SVM is proposed to improve the performance of DRC access.The WLC only consider server performance and the connection of the number of connections and does not take the actual running state of the defectinto account, from the cluster server access response time of proposed a LVS scheduling algorithm based on SVM is improved to enhance the performance of DRC access.