load balancing in cloud computing

8/10/2019 Load Balancing in Cloud Computing

1/35

NATIONAL INSTITUTE OF TECHNICAL TEACHERS TRAINING AND RESEARCH, BHOPAL (M.P.)

Performance Enhancement of Cloud

Computing using Clustering

Submitted By:

Bhupendra PanchalM.Tech (CTA)

0012CS11MT04

Guided By:

Dr. R. K. Kapoor

NITTTR Bhopal

Dept. of Computer Engineering & Application

A Dissertation

on


2/35

Agenda

Aim & Objective

1

2

Modelling the VM Allocation3

Introduction

4 Proposed Dynamic VM Allocation Algorithm using Clustering

6 Simulated Outcomes

Implementation of Proposed Algorithm in CloudSim5

7 Comparing the Results

8 Conclusion

9 Future Work


3/35

1. Introduction

Cloud computing is an emerging infrastructure paradigm that allows

efficient maintenance of cloud services.

It allows cost reduction, flexibility, scalability and reducing the power

consumption with efficient use of servers.

Cloud computing regroups the use of different technologies such as

virtualization, clustering.

This technology begins to become really popular because it allows

customers to pay only for what they are using.

If a company needs an important computing power for few days they can

buy it for few days on the cloud rather than to buy machine especially for

that purpose.


4/35

Introduction

Cloud computing can be defined as

a type of parallel and distributed system consisting of interconnected

and virtualized computers that are dynamically provisioned and

presented as one or more unified computing resources based on service-

level agreements between the service provider and consumers.

It offers different services as IaaS, SaaS, PaaS .


5/35

1.2 Architecture of Cloud Computing

Cloud computing architecture can be viewed as layeredmodel which can be divided into 4 layers: the

hardware/datacenter layer, the infrastructure layer, the

platform layer and the application layer as shown in

Figure.

This layering architecture of cloud provides the

integration of each layer and their relationships.

It helps to understand the basic idea of cloud

infrastructure that how this computing is structured. This

layering architecture is shown here:


6/35

Architecture of Cloud Computing


7/35

2. Aim and objective

The aim of this study is to improve the performance of cloud

computing environment using clustering technology.

Following are the objectives of the study in specific terms

To improve the performance of cloud computing by implementing

a new VM allocation algorithm using clustering.

Evaluate the performance of proposed algorithm and comparing it

with existing VM allocation algorithms.


8/35

The primary aim of this study is to improve the performance of cloud

computing.

Performance of any cloud environment is based on CPU utilization,

memory utilization, network bandwidth and I/O performance.

Efficient utilization of available resources among the VM can providean efficient behavior of cloud computing. This sharing of resources is

based on different scheduling and allocation policies.

Our methodology is to use clustering technology with virtual

environment. Cluster based data centers are highly efficient for

performing fast accessing of resource and I/O operations.

Aim and objective


9/35

3. Modeling the VM Allocation

The Data Center Controller uses a VmLoadBalancer to determinewhich VM should be assigned the next request for processing.

The Vmloadbalancer are :

Round Robin, throttled and active monitoring load balancer

Round Robin Load Balancer (RRLB): In this approach, the

datacenter controller assigns the requests to a list of VMs on a

rotating basis. The first request is allocated to first VM and assigns

the upcoming requests in a circular order. Once the VM is assigned

the request, the VM is moved to the end of the list.


10/35

Modeling the VM Allocation

Throttled Load Balancer (TLB):The TLB maintains a record of the

state of each virtual machine (busy/ideal). If a request arrived, TLB

sends the ID of ideal virtual machine to the data centre controller

and datacenter controller allocates the ideal virtual machine.

Active Monitoring Load Balancer (AMLB): The AMLB maintains

information about each VMs and the number of requests currently

allocated to which VM. When a request to allocate a new VM arrives,

it identifies the least loaded VM. If there are more than one, the firstidentified is selected.


11/35

4. Proposed Dynamic VM Allocation Algorithm

Mainly Clustering is the method which includes the grouping of similartype objects into one cluster and a cluster which includes the objects of

data set is chosen in order to minimize some measure of dissimilarity.

For scheduling the virtual machines, K-Means clustering algorithm is

used in this study.

In K-Means clustering, given data sets are divided into K number of

clusters. In this each object of data set belonging to one cluster. In each

cluster there may be a centroid or a cluster representative.

Select K points as the initial centroids

Repeat

Form K Clusters by assigning all points to closest centroid

Recomputed the centroid of each cluster

Until The centroids dont change

K-Means Algorithm:


12/35

Proposed Dynamic VM Allocation Algorithm

Input: List V of Virtual Machine`s with their location around the globe

List D of datacenters

Algorithm:

1. Select K points according to the number of datacenters in D

2. Choose datacenter from D

3. Form K clusters of VM`s from V by assigning closest centroid

4. Recomputed the centroid of each cluster

5. Arrange all the requested VM`s in cluster form

6. Allocate the VM`s to the available Host

7. If all the VM`s are allocated

8. Assign the VM`s cluster to the selected datacenter9. Endif

10. Repeat [2] until D is empty

11. If all the VM`s are created in the datacenters

12. Endif

13. compute the results


13/35Working flow diagram of proposed VM allocation scheme


14/35

5. Implementation of Proposed Algorithm

5.1 Implementation of CloudSim in Eclipse

CloudSim is a java based simulation tool, so it can be used either with the

eclipse IDE or NetBeans IDE. For this work, we select eclipse Juno IDE to

implement proposed VM allocation policy.

To run the CloudSim in eclipse Juno, we have to download the eclipse IDE

and install it. Since eclipse is java base platform, so a java run time

environment is needed before installing it.

When this installation is completed, latest CloudSim package is extracted

and it is imported in the eclipse.

When we import the CloudSim in eclipse, it asks for the path where the

extracted CloudSim package is available. This package is then copied into

the workspace of eclipse from where it can be easily configured and run.


15/35


16/35


17/35


18/35


19/35


20/35


21/35


22/35


23/35

Implementation of Proposed Algorithm

5.2 Implementation of Dynamic VM Allocation Algorithm

Since our concept is to allocate the virtual machines dynamically, as per

user request, So a graphical interface is prepared for selecting the

position of virtual machines.

From here lists of VM and their location are created and submitted, for

starting the simulation with CloudSim.

Number of datacenters is assumed to be known in advance. On the basis

of datacenters, clusters of VM`s are created.

In the CloudSim, DatacenterBroker class is the base class for any VM level

event. When createVmsInDatacenter (int datacenterId) function is called,

it performs the VM allocation to the respective datacenter.


24/35

6. Simulated Outcomes

In the CloudSim simulator, we have performed the test in both cases: first

with the existing allocation algorithm that is based on selecting host with

minimum Pe`s and second with the proposed Dynamic VM allocation

algorithm with clustering.

For both the cases, we set the same parameters to perform the test such

as:Number of VM`s: 15

Number of Datacenters: 3

Number of Cloudlets: 40

Number of Hosts: 3

Simulation results for both the cases are given here:


25/35

Simulated Outcomes

6.1 Test Perform with Existing VM Allocation Algorithm

Fig2: Test Perform with Existing VM Allocation Algorithm


26/35

Simulated Outcomes

6.2 Test Perform with Dynamic VM Allocation Algorithm with Clustering

Fig3: VM Allocation


27/35

Simulated Outcomes

Test Perform with Dynamic VM Allocation Algorithm with Clustering

Fig4: Output Window


28/35


29/35

7. Comparing the Results

Results obtained from both algorithms are arranged in the table :

Table1: Result comparison of both Algorithms

According to the table, overall debt for each datacenter is minimum in

the dynamic allocation policy and it also allows proper load sharing

among the datacenters.

Thus it reveals that clustering technique helps to improve the

performance and it also allows fast accessing and releasing of resources.


30/35

Comparing the Results

The comparing output with these values is represented in the graph as:

Fig6: Performance comparison of Proposed Algorithm with existingallocation Algorithm using same input parameter


31/35

8. Conclusion

As cloud computing itself is a bigger umbrella that is merged with

different technologies. This complex integration is also having a bad

impact on the performance of cloud computing.

This work was aimed to improve the performance of cloud computing

with Dynamic VM allocation Algorithm using Clustering.

It arranges virtual machines in cluster form before allocating them to the

datacenters. This arrangement provides efficient CPU utilization and load

sharing among the datacenters, so performance can be enhanced in

some aspects.

The simulation work is done with CloudSim toolkit that was implemented

on Eclipse Juno version. Simulated results are analyzed and compared

with existing methods. It is obtained from the results that proposed

algorithm is having better performance results than existing algorithm.


32/35

9. Future Work

It is reviewed in this work, that there are large gaps to be resolved in

cloud computing. This work is performed to analyzing performance

parameters and its enhancement using clustering technology.

Although integration of K-Means clustering works well than the existing

methodologies in cloud environment but this also can be replaced with

other available clustering techniques.

This K-Means algorithm is having best results for large collection of data.

So clustering technology can also be implemented at datacenter level for

Dataas a Service.

Caching is also a solution for improving the performance. It allows fast

accessing of data. In this way each cluster would have a copy of database

so it would reduce the number of hits to datacenters and hence improve

performance.

REFERENCES


33/35

[1] G. Malathy and Rm. Somasundaram, Performance Enhancement in Cloud Computing using Reservation

Cluster,European Journal of Scientific Research, ISSN 1450-216X, Vol. 86 No 3, September, 2012, pp.394-401

[2] Bhupendra Panchal and R. K. Kapoor, Performance Enhancement of Cloud Computing with Clustering,

International Journal of Engineering and Advanced Technology (IJEAT) ISSN: 2249 8958, Volume-2, Issue-5, June2013

[3] Nikolaus Huber, Marcel von Quast, Micahal Hauck and Samuel Konev, Evaluating and Modeling

Virtualization Performance Overhead for Cloud Enviornments

[4] A. Mahendiran, N. Saravanan, N. Venkata Subramanian and N. Sairam, Implementation of K-Means

Clustering in Cloud Computing Environment, Research Journal of Applied Sciences, Engineering and Technology4(10): 1391-1394, 2012, ISSN: 2040-7467

[5] Michael Shindler, Alex Wong and Adam Meyerson Fastand Accurate k-means For Large Datasets

[6] Pinal Salot, A Survey of Various Scheduling Algorithm in Cloud Computing Environment International

Journal of research and engineering Technology (IJRET), ISSN: 2319 1163, Volume: 2 Issue: 2, Feb 2013

[7] Rodrigo N. Calheiros, Rajiv Ranjan, Anton Beloglazov, Csar A. F. De Rose, and Rajkumar Buyya, CloudSim: AToolkit for Modeling and Simulation of Cloud Computing Environments and Evaluation of Resource Provisioning

Algorithms

[8] Alexandru Iosup, Simon Ostermann and M. Nezih Yigitbasi, Performance Analysis of Cloud Computing

Services for Many-Tasks Scientific Computing,IEEE Transaction on parallel and distributed system, VOL. 22, NO.

6, JUNE 2011

[9] Manisha Malhotra, Simulationfor Enhancing the Response and Processing Time of Datacenter Simulation,

REFERENCES

REFERENCES


34/35

[10] Navjot Kaur, Jaspreet Kaur Sahiwal, Navneet Kaur, EfficientK-Means Clustering Algorithm using Ranking

Method in Data Mining, International Journal of Advanced Research in Computer Engineering & Technology,

Volume 1, Issue 3, ISSN: 2278 1323, May2012

[11] Prof. S.China Venkateswarlu, Prof. M.Arya Bhanu, Prof. Yudhaveer Katta and V.Badari, Implementationof K-Means Clustering Algorithm Using Java,Global Journal of Computer Science and Technology, Volume 11, Issue

17, Version 1.0, October 2011

[12] Parul Agarwal,M. Afshar Alam and Ranjit Biswas, Issues,Challenges and Tools of Clustering Algorithms,

IJCSI International Journal of Computer Science Issues, Vol. 8, Issue 3, No. 2, ISSN: 1694-0814, May 2011

[13] Tejinder Sharma and Vijay Kumar Banga, Efficient and Enhanced Algorithm in Cloud Computing,International Journal of Soft Computing and Engineering (IJSCE) ISSN: 2231-2307, Volume-3, Issue-1, March 2013

[14] Jasmin James and Dr. Bhupendra Verma, Efficient VM Load Balancing Algorithm for a Cloud Computing

environment,International Journal on Computer Science and Engineering (IJCSE), ISSN: 0975-3397, Vol. 4 No. 09

Sep 2012

[15] Sonam Rathore, EfficientAllocation of Virtual Machine in Cloud Computing Environment , International

Journal of Computer Science and Informatics, ISSN: 2231 5292, Volume-2, Issue-3, 2012

[16] Soumya Ray and Ajanta De Sarkar, ExecutionAnalysis of Load Balancing Algorithms in Cloud Computing

Environment, International Journal on Cloud Computing: Services and Architecture (IJCCSA), Vol.2, No.5,

October 2012

[17] A. Mahendiran, N. Saravanan, N. Venkata Subramanian and N. Sairam, Implementation of K-Means

Clustering in Cloud Computing Environment, Research Journal of Applied Sciences, Engineering and Technology

4(10): 1391-1394, 2012, ISSN: 2040-7467

REFERENCES


35/35

Thank you.

load balancing in cloud computing

Documents