A Load Balancing Algorithm for Spatial Data in GridGIS
Yan Ma · Geography and Geo-Information Science · 2011
The wide application of spatial data needs an efficient framework to manage them and increase the availability of spatial data in geographical applications.The emergence of Grid computing coupled with Geographic Information Systems(GIS) provides an excellent framework:GridGIS,which supports fast spatial data retrieval and allows its users to transparently access data from anywhere at any time.GridGIS also causes spatial load unbalancing.This paper presents a new load balancing algorithm(TLB-Chord),which adopts the tree structure based on the classical Chord algorithm to improve system performance and increase the availability of spatial data.First,the relative researches of GridGIS and spatial load balancing are summarized.Secondly,the special thoughts of the TLB-Chord algorithm are given.The paper discusses how to construct the tree structure based on Chord,divides the peers in the system into three kinds:the only virtual root peer,some task managers and many normal peers,and gives each kind of peers how to work.Then,it is introduced that the TLB-Chord algorithm has three levels to implement the spatial load balancing:the basic level composed by normal peers,the middle level including task managers and the high level having the root peer.The algorithm begins from the basic level.Once it fails to adjust the load,the algorithm will perform the middle level.If the middle level also fails,the algorithm will adopts the high level.Thirdly,a spatial load balancing simulation system is given and the TLB-algorithm(peers connected in Chord based on the tree structure) and the common algorithm(peers connected in the physical order based on a ring structure) are compared through the testing environment in GridGIS,which uses the iterative way,and shows the TLB-Chord algorithm can improve the system performance better.