Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Computer network > General issues > Theory of computer networks

Research and Implementation of Dynamic Distributed Strategy Load Balancing of Grid Service

Author PengYunYa
Tutor LuoSiWei
School Beijing Jiaotong University
Course Computer System Architecture
Keywords Grid MDS Dynamic Load Balancing Distributed Scheduling Strategy Multiserver Load Balancing
CLC TP393.01
Type Master's thesis
Year 2008
Downloads 172
Quotes 1
Download Dissertation

Storage grid which can rapidly uniformly view and manage storage data, also provides more data protection. Storage grid with these characteristics, makes storage grid technology widely considered that it is one of the most extensive technologies in storage field.This paper is a research for campus network storage grid application. Taking into account the practical application environment of the campus network, this paper basing on the DNS principle realizes dynamic load balancing technology. One domain name maps several web servers IP, so the requests of the domain name can be distributed to several web servers. Different web servers are charged with large number of request tasks, achieving load balancing.This paper basing on the DNS principle and Grid Information Service technology realizes dynamic distributed strategy load balancing, shortening the system response time, raising the system throughput.The static round robin load balancing algorithm can not distinguish between different servers’ performance, can not distribute more requests to the better performance server.Dynamic distributed strategy load balancing technology implements dynamic parametric linear weighted load balancing algorithm to overcome the defects of the static round robin load balancing algorithm. Furthermore, this technology realizes distributed scheduling in which each node’s monitor (MDS) provides this node’s load information. So the load balancer doesn’t need to collect servers’ load information initiatively, thereby, reducing the extra overhead causing by the collection of load information, lightening the load balancer’s burden, reducing the risk of load balancer being a system bottleneck.

Related Dissertations
More Dissertations