Tanveer Ahmed, Yogendra Singh/ International Journal of Engineering Research and Applications (IJERA)

ISSN: 2248-9622 www.ijera.com

Vol. 2, Issue 2,Mar-Apr 2012, pp.1027-1030
Analytic Study Of Load Balancing Techniques Using Tool Cloud Analyst.

Tanveer Ahmed Yogendra Singh

University School of Information Technology University School of Information Technology
M.Tech CSE,GGSIPU, Dwarka M.Tech CSE, GGSIPU, Dwarka
New Delhi, India New Delhi, India

Abstract
Cloud computing, as of today has boomed the use of internet computing that will give a better idea in terms of overall
services and has enhanced the use of network services where response time, cost, and datacenter service requesting time.
the capability of one node can be utilized by other nodes. 2. algorithms used
Central to the architecture is SoA and IS which raises a big
question on performance and complexity. To properly manage 2.1 Round Robin
the resources of the service provider we require balancing the It is one of the simplest scheduling techniques that utilize the
load of the jobs that are submitted to the service provider. Load principle of time slices. Here the time is divided into multiple
balancing is required as we don’t want one centralized server’s slices and each node is given a particular time slice or time
performance to be degraded. A lot of algorithms have been interval i.e. it utilizes the principle of time scheduling. Each
proposed to do this task. In this paper we present a comparison node is given a quantum and in this quantum the node will
of various policies utilized for load balancing using a tool perform its operations. The resources of the service provider
called cloud analyst. are provided to the requesting client on the basis of this time
slice. Though the algorithm is very simple but there is an
Keywords - Cloud Computing, Load balancing, Cloud additional load on the scheduler to decide the size of quantum.
Analyst, CloudSim, Virtual Machine. It is shown in fig. 1.

1. INTRODUCTION Client 2
Cloud computing [1] is service which has allowed users to Client 3
utilize the potential of other nodes on the internet and then use Client 1
the same node’s services to perform the desired task. It is
usually composed of both the services delivered by the service
provider and the datacenter that provides the services. Cloud
Service
computing provides all the features of grid computing, Provider
software as a service and utility computing. It utilizes the
concept of virtualization. Cloud computing architectures are
inherently parallel, distributed and serve the needs of multiple Client 4
Client 5
clients in different scenarios. It allows an organization to focus
on paying for the services they require rather than utilizing
traditional software services in which a fully fledged system
must be built. In [2], cloud computing is thought as of dealing
with the three issues, we here in the paper focus on the Figure1. Round Robin Algorithm
property of decision. To perform better execution of the
operations for the nodes one needs additional attributes for 2.2 Equally spread current execution load
performance calculation. These additional attributes are bound It is spread spectrum technique in which the load balancer
to increase the issue of complexity. The major issues spread the load of the job in hand into multiple virtual
associated with this paradigm are of security, complexity and machines. The load balancer maintains a queue of the jobs that
reliability. Load balancing is the mechanism that decides need to use and are currently using the services of the virtual
which requesting nodes/client will use the virtual machine and machine. The balancer then continuously scans this queue and
which requesting machines will be put on hold. Load the list of virtual machines. If there is a VM available that can
balancing can be done individually as well as on grouped handle request of the node/client, the VM is allocated to that
basis. Load balancing is also required to minimize the cost of request. If however there is a VM that is free and there is
machine and maximize the profit for the service being offered. another VM that needs to be freed of the load, then the
Hence with these issues in mind we here in this paper propose balancer distributes some of the tasks of that VM to the free
a systematic study of the various algorithms used in cloud one so as to reduce the overhead of the former VM. Figure2.
better explains the working of the ESCE algorithm. The jobs
are submitted to the VM manager, the load also maintains a
1027 | P a g e
list of the jobs, their size and the resources requested. The round robin, equally spread current execution load (ESCEL),
balancer selects the job that matches the criteria for execution and throttled. These load balancing techniques are used by the
at the present time. Though there algorithm offers better VmLoadBalancer component of the cloud analyst. These
results as shown in further section, it however requires a lot of techniques determine the order in which the application must
computational overhead. be deployed on the data centers. The default technique used by
cloud analyst is Round robin algorithm; this algorithm simply
Client1 Client 4 allocates incoming requests in a round robin fashion so this
Client 2 Client 3
algorithm does not take into account the current load of each
virtual machine. The second policy used is equally spread
current execution as the name specifies this policy equally
Load
distributes the load on each VM. And the third one is throttled,
Job Pool Balancer in this policy a threshold is defined for assigning request to
each VMs, if the number of requests increase beyond this
threshold value in all available virtual machines then the
Job Attributes requests are queued until a VM becomes available.
Cloud Analyst[3] is a GUI based tool that is developed on
CloudSim[4]architecture[5]. CloudSim is a toolkit that allows
doing modeling, simulation and other experimentation. The
Virtual main problem with CloudSim is that all the work need to be
Virtual done programmatically. The cloud analyst tool removes all the
Machine n
Machine 1
complexities by making GUI so that focus can be done on
simulation rather than programming. It allows the user to do
repeated simulations with slight change in parameters very
easily and quickly. The cloud analyst allows setting location of
Figure2. ESCE Algorithm users that are generating the application and also the location
of the data centers. In this various configuration parameters
2.3 Throttled Load balancing can be set like number of users, number of request generated
In this algorithm the client first requests the load balancer to per user per hour , number of virtual machines, number of
find a suitable Virtual Machine to perform the required processors, amount of storage, network bandwidth and other
operation. It is shown in fig.3.The process first starts by necessary parameters. Based on the parameters the tool
maintaing a list of all the VMs each row is individually computes the simulation result and shows them in graphical
indexed to speed up the lookup process. If a match is found on form. The result includes response time, processing time, cost
the basis of size and availability of the machine, then the load etc .By performing various simulations operation the cloud
balancer accepts the request of the client and allocates that provider can determine the best way to allocate resources,
VM to the client. If, however there is no VM available that based on request which data center to be selected and can
matches the criteria then the load balancer returns -1 and the optimize cost for providing services. The various activities
request is queued. performed in cloud analyst tool are summarized below

Figure4.Activities done in cloud analyst.

Figure3. Throttled Load Balancing Algorithm The main components of cloud analyst tool are:
3. CLOUD ANALYST
In order to study the various load balancing techniques we are
using cloud analyst in this three techniques are given that are
1028 | P a g e
GUI Package. It provides various user interfaces to configure tool need to be done. We have set the parameters for the user
the various simulation parameters in a better and easy way. base configuration, application deployment configuration, and
The GUI of cloud analyst is shown in figure5. data center configuration as shown in figure 6, figure 7 and
figure 8 respectively. As shown in figure the location of user
bases has been defined in six different regions of the world.
We have taken two data center to handle the request of these
users. One data center is located in is located in region 0 and
another in 2. On DC1 number of VM allocated are while that
on DC2 are 50.

Figure5.GUI of cloud analyst.

Simulation. Based on the various parameters set it executes Figure6.User Base Configuration
the simulation to provide necessary results.

UserBase. In this user base is modeled to represent the users

who deploy application.

DataCenterController. This component is used to control the

various data center activities.

InternetCharacteristics. In this component various internet Figure7Application Deployment Configuration

characteristics are modeled simulation, which includes the
amount of latency and bandwidth need to be assigned between
regions, the amount of traffic, and current performance level
information for the data centers.

VmLoadBalancer. The responsibility of this component is to

allocate the load on various data centers according to the
request generated by users. One of the three given policies can
be selected. The given policies are round robin algorithm,
equally spread current execution load, and throttled.
Figure8. Data center Configuration
CloudAppServiceBroker. The responsibility of this
component is to model the service brokers that handle traffic 5. Results
routing between user bases and data centers. The service After performing the simulation the result computed by cloud
broker can use one of the routing policies from the given three analyst is as shown in the following figures. We have used the
policies which are closest data center, optimize response time above defined configuration for each load balancing policy
and reconfigure dynamically with load. The closest data center one by one and depending on that the result calculated for the
routes the traffic to the closest data center in terms of network metrics like response time, request processing time and cost in
latency from the source user base. The reconfigure fulfilling the request has been shown.
dynamically with load routing policy works in the sense that
whenever the performance of particular data center degrades 5.1. Response Time:
below a given threshold value then the load of that data center Response time for each user base and overall response time
is equally distributed among other data centers. calculated by the cloud analyst for each loading policy has
been shown in the figure 9, 10 and 11 respectively. As can be
4. Simulation Configuration seen from the figure the overall response time of Round Robin
In order to analyze various load balancing policies policy and ESCEL policy is almost same while that of
configuration of the various component of the cloud analyst
1029 | P a g e
Throttled policy is very much low as compared to other two Figure13. Request Servicing time using ESCEL Policy.
policies.

Figure14. Request Servicing time using Throttled Policy.

5.3. Processing Cost:

The processing cost for each load balancing policy computed
by the cloud analyst as can be seen from the figures 15, 16 and
Figure9. Response time using Round Robin Policy. 17 is same so, load balancing technique does not play any role
in the processing cost.

Figure15. Processing cost using Round Robin

Figure10. Response time using ESCEL Policy.

Figure16 Processing cost using ESCEL

Figure17. Processing cost using Throttled

Figure11. Response time using Throttled Policy.

References
[1] Anthony T.Velte, Toby J.Velte, Robert Elsenpeter,
Cloud Computing A Practical Approach, TATA
5.2. Data Center Request Servicing Time:
McGRAW-HILL Edition 2010.
Data Center Request Servicing Time for each data center
[2] A Survey on Open-source Cloud Computing Solutions
calculated by the cloud analyst for each loading policy has
Patrícia Takako Endo, Glauco Estácio Gonçalves, Judith
been shown in the figure 12, 13 and 14 respectively. As can be
Kelner
seen from the figure the servicing time of Round Robin policy
[3] Cloud Analyst: A Cloud-Sim-based Tool for Modeling
and ESCEL policy is almost same while that of Throttled
and Analysis of Large Scale Cloud Computing
policy is very much low as compared to other two policies.
Environments. MEDC Project Report Bhathiya
Wickremasinghe.
[4] R. Buyya, R. Ranjan, and R. N. Calheiros, “Modeling
and Simulation of Scalable Cloud Computing
Environments and the CloudSim Toolkit: Challenges and
Opportunities,” Proc. of the 7th High Performance
Computing and Simulation Conference (HPCS 09),
IEEE Computer Society, June 2009.
Figure12. Request Servicing time using Round Robin Policy. [5] http://www.cloudbus.org/cloudsim

1030 | P a g e