Research on a Method of Geographical Information Service Load Balancing
ISBN 978-85-88783-11-9
Authors
1Heyuan, L.; 2Gang, C.
1INSTITUTE OF GEOSPATIAL,INFORMATION ENGINEERING UNIVERSITY Email: 2625845532@qq.com
2INSTITUTE OF GEOSPATIAL,INFORMATION ENGINEERING UNIVERSITY Email: leeheyuan@163.com
Abstract
With the development of geographical information service technologies, how to achieve the intelligent scheduling and high concurrent access of geographical information service resources based on load balancing is a focal point of current study. This paper presents an algorithm of dynamic load balancing. In the algorithm, types of geographical information service are matched with the corresponding server group, then the RED algorithm is combined with the method of double threshold effectively to judge the load state of serve node, finally the service is scheduled based on weighted probabilistic in a certain period. At the last, an experiment system is built based on cluster server, which proves the effectiveness of the method presented in this paper. 1 Introduction Geographical information is playing an increasingly important role on government management decision-making, national defense safety and people's living standards improve. Along with our country economy and the advancement of national defense informatization construction, agencies at all levels and the public’s demand to authority, reliable geographic information service increases, need to implement comprehensive utilization and online services of multi-scale, multi-type of geographic information resources[1][2].The current network geographic information service system is facing the unpredictable challenge of concurrency growth, capacity limits and system response, and geographic information service scalability and availability is being more and more attention[3].Geographic information service dynamic load balancing is to solve how to make virtual multiple service node based on the Web server cluster into a logically unified "super" geographic information service center, implement the virtualization management and intelligent scheduling for each service node within the cluster resources, in order to support the service of dynamic binding, find and replace, thus improve the concurrent access ability of cluster system. Such as amazon services can be on-demand intelligent scheduling system capacity (such as servers, storage, and network bandwidth), can be flexible deployment a variety of services from provider resources, without the need for extra configuration for the uncertain demand in advance[4]. Among existing network Map service systems, the Google Map has the intelligent scheduling ability of geographic information service resources with the support of dynamic load balance. In the field of other Web services, common load balance algorithm are rotation scheduling algorithm (RR) and local perception request distribution algorithm (LARD). The main problems of the RR algorithm is not considering the situation of load each service node, resulting in a decline in performance of the system. If a certain type of service request rate is high, the algorithm of LARD leads to a node of the utilization rate is very high, and the other nodes are idle for a long time, wasting server resources[5] [6]. In this paper, we proposed a dynamic load balancing algorithm for multiple service node. According to service node in the cluster, using designed load balancing algorithm to share a large number of concurrent access to business to multiple processing nodes respectively, reducing the time for a response, so as to achieve more service node's geographical information service virtualization and global load balancing.
Keywords
geographicinformation service; load balancing; intelligent scheduling