Implementation of the load balancing algorithm over a local area network and the internet

Abstract

In this paper, experimental evaluation of the load balancing algorithm in real environments is presented. we emphasize on the effects of delays in the exchange of information among nodes, and the constraints these effects impose on the design of a load balancing strategy. two testbeds in two different real environments have been built; The first implementation was over a local area network whereas the second one was over Planet-Lab. The results show the effect of the network delays and variances in the tasks processing time on choosing adequate gain values for the load balancing algorithm.

Distributed Load Balancing in the Presence of Node Failure and Network Delays

“…The second and third steps employ the method introduced in (Chiasson et al, 2005;Ghanem et al, 2004a) and used earlier in the literature in the Sender/Receiver Initiated Diffusion (SID and RID) policies (Saletore, 1990;Willebeek-LeMair and Reeves, 1993) to calculate the fractions p ij . This method is attractive since it only considers nodes that have queue sizes less than the average, and therefore results in as few connections as possible when the excess load is moved out of node j which makes the policy at hand more scalable.…”

“…It considers deterministic communication and transfer delays. Experimental evaluations of the model using its derived policy were conducted over a LAN and the Internet where optimization over the gain parameter K was performed (Ghanem et al, 2004a;Chiasson et al, 2005). The limitations and the overheads involved with the implementation of the deterministic-delay load-balancing policy in a random delay environment has been discussed by the authors in .…”

An Adaptive Load Sharing Algorithm for Heterogeneous Distributed System

“…The distributed system consists of several nodes and the same load balancing software is installed to run on all the nodes in the distributed system. By installing the same software in all the nodes, the load balancing decision is taken by a node locally (decentralized) by collecting the information from the neighboring nodes as opposed to the centralized load balancing policy [14].…”

“…The node status information is responsible for disseminating the load information to the node that has requested it. The exchange of the information has a profound effect on the load balancing decision; it has to be done according to the predefined intervals of time specified on each node [7] [14].…”

“…UDP is preferable as it incurs less communication overhead. In general the architecture provides three 10 P. Neelakantan www.ijorcs.org layers: Communication layer, Load balancing process and application layer [14][10]. For storing information two data structures were used.…”