An Active Approach for Load Balancing in Grid

International Journals of Advanced Research in
Computer Science and Software Engineering
ISSN: 2277-128X (Volume-7, Issue-6)
Research Article
June
2017
An Active Approach for Load Balancing in Grid Computing
Dr. B. Mahesh
Department of CSE, Malla Reddy Engineering College & Management Sciences,
Medchal, Telangana, India
Abstract—Grid is connected by parallel nodes that form a computer cluster, which runs on an operating system,
Linux or free software. The cluster can differ in size from a small work station to several networks. The technology is
efficient to a wide range of applications, such as mathematical, scientific or instructive tasks through some computing
resources. Grid computing is a processor architecture that combines computer systems from various domains to attain
a main objective. In grid computing, the computers on the network can work on a mission collectively, thus operation
as a supercomputer. Load balancing strategies are divided into two categories. Each category is considered base on
the origin of grid resource topology that includes the flat resource topology and hierarchical resource topology.
However for computational grids we must pact with main new issue like: heterogeneity, scalability and adaptability. I
suggested a layered algorithm which achieves active load balancing in grid computing. Based on a tree model, an
algorithm presents the following features: (i) it is layered; (ii) it supports heterogeneity and scalability; and, (iii) it is
totally independent from any physical architecture of a grid.
Keywords— active load balancing, grid computing, layered algorithms, tree model, physical architecture
I. INTRODUCTION
The improvement of performance of computers and their cost reduction is due to development in computing resources.
The modern researches on computing architectures has resulted the personification of a new computing paradigm known
as Grid computing. Grid is a kind of disseminated system which maintains the partaking and Synchronized use of
geologically spotted and multi owner resources autonomously from their substantial type and location, in active implicit
organizations that share the same objective of solving large-scale applications. The main aim is to prevent the condition
where some processors are overloaded with a set of tasks while others are evenly loaded or even idle. The load balancing
of the jobs in a grid environment can significantly control grid’s performance as the resources are active in nature. Hence
the load of resources changes with difference in configuration of grid. A key attribute of Grids is that resources like CPU
cycles and network capacities are shared among various applications, and then, the amount of resources available to any
given application highly fluctuates over time. Load balancing uses parallelism for boosting throughput in order to reduce
the response time by brightly distributing application resources. Load balancing can be implemented using two
techniques i.e. Static load balancing and dynamic load balancing. Static load balancing algorithms assign the jobs of a
parallel program to workstations based on moreover the load at the time nodes are allocated to some task, or based on an
average load of our workstation cluster. Dynamic load balancing algorithms alter sharing of work during run time. Dynamic
load balancing algorithms use the efficient current information of the load for taking decisions about load distribution.
II. GRID STRUCTURE
The uncomplicated grid can be defined as an unified system for a distribution of non interactive workloads that involve a
fat number of files. Grid can be built in all sizes range from just a small number of machines to the cluster of machines
structured as a hierarchy spanning the world. The Simplest Grid consist of just a small amount of machines all the same
hardware architecture and same operating system connected on a local network. Because the machines have the same
architecture choosing application software for these machines is usually simple. This type of structure the grid is known
as intergrid which is homogeneous systems. When the machines are been include to the heterogeneous systems, more resources
are available. File partaking may still be accomplished using network file systems. Such a grid is referred as an intragrid.
Fig 1: Simple Grid Architecure
© www.ijarcsse.com, All Rights Reserved
Page | 514
Mahesh International Journals of Advanced Research in Computer Science and Software Engineering
ISSN: 2277-128X (Volume-7, Issue-6)
In specifying the various layers of the Grid architecture, we follow the principles of the ―hourglass model‖. The thin neck
of the hourglass define a petite set of core abstractions and protocols (e.g., TCP and HTTP in the Internet), onto which
numerous different high-level behaviours can be mapped (the top of the hourglass), and which themselves can be mapped
onto numerous different underlying technologies (the base of the hourglass). By definition, the amount of protocols
defined at the neck must be small. In our architecture, the neck of the hourglass consists of Resource and Connectivity
protocols, which make possible the partaking of individual resources. Protocols at these layers are designed so that they
can be implemented on top of a assorted range of resource types, defined at the Fabric layer, and can in turn be used to
build a wide range of global services and application-specific behaviours at the Collective layer—so called because they
involve the synchronized (―collective‖) use of multiple resources. Our architectural explanation is high level and places
few constraints on design and implementation.
Fig 2: Layered Grid Architecture
III. TREE BASED BALANCING MODEL
In order to well explain the proposal model, we must define the topological structure of a Grid computing. Mapping a
Grid into a tree-based model: The load balancing strategy proposed in this is based on a mapping of any Grid into a tree
based model. It is build as follows:
 First, for each site we create a two levels sub tree. The leaves of this sub tree correspond to the Computing
Elements of a site and the root of this sub tree is a virtual node associated to the site. The role of this virtual
node is to manage the workload of a site. In practice, this management function is processed by a computing
element within the site.
 Second, the subtrees corresponding to all sites of a cluster are aggregated to generate a three levels sub tree.
 Third, these subtrees are connected together to build a four levels tree.
Level 0: In this first level (top level), we have a virtual node that corresponds to the root of the tree. It is associated to
the Grid and it manages the workload on the whole Grid.
Level 1: This level contains G virtual nodes, each one associated to a substantial cluster of the Grid. In our load
balancing strategy, this virtual node is responsible to manage its sites.
Fig 3: Tree based representation of Grid
Level 2: In this third level, we find S nodes associated to physical sites of all clusters of the Grid. The main function of
these nodes is to manage the workload of their physical Computing Elements.
© www.ijarcsse.com, All Rights Reserved
Page | 515
Mahesh International Journals of Advanced Research in Computer Science and Software Engineering
ISSN: 2277-128X (Volume-7, Issue-6)
Level 3: At this last level (leaves of the tree), we find the M Computing Elements of a Grid linked to their Respective
and clusters. The final tree is denoted by G/S/M , where G is the number of Clusters that compose the Grid, S the number
of Sites and M the number of CE’s. As This generic tree can be transformed in turn into three specific trees: G/S/M,
1/S/M and1/1/M, depending on the values of G, S and M. The mapping function generates a non cyclic connected graph
where each level has specific functions.
IV. LOAD BALANCING ALGORITHMS
We define three levels of load balancing algorithms:
Intra-site load balancing algorithm: This algorithm is considered as the kernel of our load balancing strategy. It is
executed when CE’s managers find that there exists an imbalance between computer elements under their control. To
make this report, the managers receive periodically local workload information from computing elements. Based on this
information and on the Avrg of average grid workload, the managers analyze the workload of sites periodically. Depending on the result of this analysis, either they decide to start a local load balancing between CE’s of the same site, or
either they inform their manager site (cluster) that are currently overloaded.
BEGIN
For Every CEijk AND Periodically do
-Update actual workload Lijk of CEijk
-Send load information to CE’s manager Sjk
end For
- Update current load Ljk of site Sjk
- Send load information to sites manager Ck
- Receive grid average load from Ck
If((Ljk−Avrg)/Ljk>T) then
imbalance state
else
return
end If
Intra-cluster load balancing algorithm: This algorithm is executed only when some CE’s managers fail to balance
locally the overload of their CE’s. Knowing the global state of each site, the sites manager can evenly distribute the
global overload between its sites.
BEGIN
For Every site Sjk of Ck AND Periodically do
Update current workload Ljk and send it to sites
Manager Ck
end For
- Update actual load Lk of Ck
- Send it to grid manager
- Receive grid average load from grid manager
If(Overloaded sites number≥sit-max) then
Ck is in imbalance state
else
Return
end If
[Partition sites of Ck into overloaded, underloaded and balanced sites]
SS←∅; SR←∅; SN←∅
Intra-grid load balancing algorithm: This third algorithm performs a global load balancing among all clusters of a grid.
It is executed only if the other two levels are failed to achieve a complete load balance. It is significant to remark that at
this level, the algorithm always succeeds load balancing all these clusters. It is thus useless to test if some clusters are
still imbalanced.
BEGIN
For EveryCk AND Periodically do
- Update current workload Lk
- Send it to grid manager
end For
- Update global grid workload
- Compute grid average load
- Send it to all clusters
If(overloaded clusters number≥clu-max) then
© www.ijarcsse.com, All Rights Reserved
Page | 516
Mahesh International Journals of Advanced Research in Computer Science and Software Engineering
ISSN: 2277-128X (Volume-7, Issue-6)
Grid is imbalanced
else
Return
end If
[Grid Partition into overloaded, underloaded and balanced clusters]
CS←∅;CR←∅;CN←∅
V. CONCLUSION AND FUTURE WORK
In this study, we addressed the problem of load balancing in large scale disseminated systems. We proposed a load
balancing strategy based on a tree representation of a Grid. The model allows transforming any Grid architecture into a
unique tree with at most four levels. From this generic tree, we can derive three sub-models depending on the elements
that compose a Grid. Using this model, we defined a hierarchical load balancing strategy that privileges local balancing
in first (load balance within sites without communication between sites). The first results of our experimentations are
very promising and lead to a better load balancing between CE’s of a Grid without high computing overhead. We have
appreciably improved the metrics defined, in particular average response time. This experimental result also shows that
the proposed algorithm enhanced the load balancing and for future work. It can introduce more effectively with resource
utilization using better scheduling approach.
REFERENCES
[1]
Albert Y. Zomaya, Yee-Hwei Teh, "Observation on Using Genetic Algorithms for Dynamic Load Balancing",
IEEE, Volume 12, Issue 9, September 2001.
[2]
Roy D. Williams, "Performance of Dynamic Load Balancing Algorithms for Unstructured Mesh Calculations",
Concurrent Supercomputing facilities, June 1990
[3]
Buyya, R., D.Abramson, J. Giddy and H.Stockinger, 2002. Economic models for resourcemanagement and
scheduling in grid computing. J.Concurrency and Computation: Practice and Experience, 14: 1507-1542.
[4]
E. Deelman A.Chervenak and al. High performance remote access to climate simulation data: a challenge
problem for data grid technologies. In Proceeding. of 22th parallel computing , volume 29(10), pages 13–35,
1997.
[5]
E. Badidi.Architecture and services for load balancing in object disseminated systems . PhD thesis, Faculty of
High Studies, University of Montreal, Mai 2000.
[6]
F. Berman, G. Fox, and Y. Hey. Grid Computing: Making the Global Infrastructure a Reality . Wiley Series in
Communications Networking & Disseminated Systems, 2003.
[7]
T.L. Casavant and J.G. Kuhl. A taxonomy of scheduling in general purpose disseminated computing systems.
IEEE Transactions on Software Engineering , 14(2):141–153, 1994.
[8]
[Kaur& Singh, 2012] PawandeepKaur, Harshpreet Singh, ―Adaptive dynamic load balancing in grid computing
an approach,‖ International journal of engineering science & advanced technology, ISSN: 2250–3676
Volume-2, Issue-3, 625 –632, May-Jun 2012.
[9]
[Srivastava, 2011] Prabhat Srivastava, ―Improving Performance in Load Balancing Problem on the Grid
Computing System‖, International Journal of Computer Applications (0975–8887), Volume 16–No.1, February
2011
© www.ijarcsse.com, All Rights Reserved
Page | 517