Study On Deploing Technologies Of Geoinformation Server Cluster Based On Nginx

With widely application of Geo-Information Service, the network accessing and flow hasshown expolsive growth trend that bing great pressure and seriously affects of popularizationand application.Therefore, the research how to improve the concurrent users and responsetime of Geo-Information Service has important theretical and practical significance topromote the application and popularization of Geo-Information Service.At present, the cluster is the mainstream to improve the service quality when mass clientssimultaneously access。In the computer field, most sites are implemented by software, whileit is still few in the Geo-Information Servives files. Therefor,the paper depath study relatedtechnologies of build high perdormance of Geo-Information Servives platform based on theopen source software of Nginx,the main work and results are as follows.(1) Discussed the concept and characteristics of computer cluster, the operationalprinciple and realization way and classification of load balancing cluster; and comparated theload balancing schedulings.(2) Analysis the structure and working flow of Nginx. In Linux, the Geo-InformationService cluster is bulided by softwares of Nginx and Geoserver, and detaily explained theconfiguration items by Beijing city and GanZhou areas. The single service computer and RRand WLC are tested by the LoadRunner, proved the validity of clustering technology.(3) Aiming at the drawbacks of WLC and the characteristics of Geo-InformationServives, the paper design the algorithms of the adaptive scheduling algorithm based ondynamic feedback, and implemented in Nginx.In the design process, the load constrainted byphysical properties of server get the integrated load rate LOAD which can more accuratelyreflect the performance of different server has same load;according to the update weight isnonzero,get―available server array‖and―not available server array‖,the scheduling moduledirectly calls available server optimization measures to improve the efficiency of the system.(4) Load testing of WLC and the proposed algorithm, test results show that the proposedalgorithm is lower8seconds in the average response time, and the proposed algorithm ismore20in the number of mass clients simultaneously access, that fully proved the validity ofthe proposed algorithm.