ISSN: 2393-8528
Contents lists available at www.ijicse.in
International Journal of Innovative Computer Science & Engineering
Volume 2 Issue 3; July-August-2015; Page No. 44-51
Page44
Implementation of Randomized Hydrodynamic Load Balancing Algorithm using Map
Reduce framework on Open Source Platform of Hadoop
Rashi Saxena
1
, Sarvesh Singh
2
1
(Research Scholar, Jayoti Vidyapeeth Women’s University, India)
2
(HOD, Computer Science & Engineering Department, Jayoti Vidyapeeth Women’s University, India)
ARTICLE INFO ABSTRACT
Received: 23 June 2015
Accepted 18 July 2015
Corresponding Author:
Rashi Saxena
Department of Computer Science
and Engineering, Jayoti Vidyapeeth
Women’s University, Jaipur
Load balancing is performed to achieve the optimal use of the existing
computational resources as much as possible whereby none of the resources
remains idle while some other resources are being utilized. Balanced load
distribution can be achieved by the immigration of the load from the source
nodes which have surplus workload to the comparatively lightly loaded
destination nodes. Applying load balancing during run time is called dynamic
load balancing (DLB). This paper presents the randomized hydrodynamic
load balancing (RHLB) method which is a hybrid method that takes
advantage of both direct and iterative methods. Using random load
migration as a direct method, RHLB approach intends to solve the problems
derived from the exceptional instantaneous load rises, and diffuse the
surplus workload to relatively free resources. Besides, using hydrodynamic
approach as an iterative method, RHLB aims to consume minimum possible
system resources to balance the common workload distributions. The results
of the experiments designate that, RHLB outruns other iterative based
methods in terms of both balance quality and the total time of the load
balancing process.
© IJICSE, All Right Reserved.
Key words: Dynamic Load
Balancing, Randomized Hydrodynamic
Load Balancing.
INTRODUCTION
High performance computing (HPC) systems are very
effective at solving problems that can be partitioned
into small tasks. Distributing tasks among computing
nodes (CNs) is the key issue to increase the throughput
and the utilization of an HPC system. Any strategy, used
for even load distribution among CNs, is called load
balancing (LB). The main purpose of load balancing is to
keep all computational resources in use as much as
possible, and not to leave any resource in idle state
while some other resources are being utilized.
Conceptually, a load balancing algorithm implements a
mapping function between the tasks and CNs. Load
balancing is especially more crucial in case of the no
uniform computational patterns and/or in
heterogeneous environments. In dynamic load
balancing (DLB), the load is balanced during the
computation process. The primary mechanism of the
DLB is the notion of migration of tasks among CNs.
Given the instantaneous load distribution of the whole
system, queued tasks, waiting to be executed, are
migrated from a heavy loaded CN to a relatively free
CN. Thus, even in case of non-uniform computational
patterns or instantaneous computational power needs,
balanced workload distribution among the whole
system can be achieved and the utilization of the CNs
can be increased. However, any DLB method has to deal
with some issues like forecasting the right time to
invoke the balancing process, selection of the node that
makes the load balancing decisions and how to migrate
the load among the CNs. Dynamic load balancing
techniques can be classified according to migration
operation as direct and iterative methods. Direct
methods aim to find out the last execution node of the
surplus load in one load balancing step. Hence, direct
methods need actual information about the load
distribution of the whole system. These kinds of
methods are most suitable for systems equipped with a
broadcast mechanism [2]. Load balancing process is
performed by migration of the partition objects that
contain the surplus workload. Load transfer is
controlled by two dynamically selected threshold
values. The process is invoked by the sender (highly
loaded) node and the invocation mechanism is