Energy Aware Of Cloud Computing Development Using Cloud Simulators

Abstract

Keywords: Cloud SimulatorsCloudSimGreenCloudCloud AnalystEMUSIMGroudSim

Introduction

Although cloud computing started to be used in production more than a decade ago, we still have

many challenging situations that requires a significant amount of research to be done. A physical cloud

for conducting research it is not practical for medium to small sized educational institutions. A physical cloud requires a considerable investment in necessary equipment (servers, networking

devices, air conditioning). A solution for this situation is to use cloud simulators which can create a

virtual cloud infrastructure without the considerable investments of a real cloud.

A cloud simulator assists at creating various model of cloud applications by making data centers,

virtual machines and other utilities that can be configured appropriately, thus making it easier to analyze.

Until today many cloud simulators had been developed and have been actively used to conduct cloud

research. These simulators have different features like GUI, base programming languages, extensibility

through plugins or modules, etc.

The academic websites and journals includes now many articles comparing different cloud

solutions. This article takes a deeper approach on cloud simulators. In this article I present a

comprehensive study of multiple cloud simulators by highlighting their features and analyzing advantages

and disadvantages. A conclusion with the best cloud simulator for a certain project is provided at the end

of the article.

A comparative analysis of various cloud simulators is presented on table 1 .

Benefits of using cloud simulators over physical cloud are as follows:

-Low Cost : The cost of a cloud simulator software is much less when compared to purchasing hardware and proprietary software (OS, VM and cloud software, etc). Many

simulators are available free of charge this making cloud simulators more convenient. Free cloud

simulators are usually also open source. This opens the possibility of developing custom features

that haven’t been designed in the original software.

- Repeatable and Controllable : Our test results are not affected in any way by hardware limitations, background load of other processes or network speed as in real world. This permits

developing of the desired model without any other variables involved. The model can be tested

as many times we want, until we get the desirable output.

- Environment : A simulator provides environment for testing of various scenarios under different workloads.

Cloud Simulators

This section takes in consideration various cloud simulators that are mostly used to conduct cloud

research simulations. We have described in this section 6 cloud simulators.

2.1 CloudSim

CloudSim is one the most popular simulation tool available for cloud computing environment.

CloudSim it is a library for simulation of Cloud computing environments. Software has been developed at

the Computer Science and Engineering Department of the University of Melbourne, Australia at

CLOUDS Laboratory. The classes provided within library permits describing data centers, virtual

machines (VM), applications, computational resources, and policies necessary for management of

different parts of the system. The simulation tool can be adapted to our requirements by simply extending

or replacing the classes built in. As we have mentioned before, CloudSim is not a ready to use solution

where we set parameters and gather results necessary to use for our project. Being a library, CloudSim requires that we write a Java application using its components following our desired scenario and then

collect the obtained results for the analysing the performance of Cloud applications. (“CloudSim

Simulation Framework - Superwits Academy”, 2016) (Umang, S. & Ayushi ,S. 2016).

The layered architecture of CloudSim is outlined on Fig. 1 .

Cloudsim contains the following features:

Support for modelling and simulation of large scale Cloud computing systems

Permits modelling and simulation of virtualized server hosts

Support for simulation of network connections together with the simulated system components

Support for simulation of federated cloud environment that includes also inter-network resources

for both private and public domains

Permits user-defined policies for allocation of hosts to virtual machines and policies for

allocation of host resources to virtual machines

Support for user-defined policies at allocation of hosts and host resources to virtual machines

Permits simulation of electric energy consumption for desired model

Support for modelling and simulation of application modules

permits insertion of simulation elements and also stop and resume of simulation

Open source

See Full Size >

2.2 Green Cloud

GreenCloud is a packet level simulator that has been created as an extension to well-known

network simulator NS2. GreenCloud has been developed in the context of ECO‑CLOUD and GreenIT

projects. This simulator is specially made for energy-aware environment. GreenCloud permits calculation

of energy consumption for any particular cloud computing component such as servers, network switches

and communication links. (Boru, D. & Kliazovich, D. & Granelli, F. & Bouvry, P. & Zomaya,

A.Y.,2015)

GreenCloud features:

Permits simulation of electric energy consumption for desired model

Defining of energy models for each type of cloud component

Simulation of CPU, memory, storage and networking resources with special attention to

networking component

Supports virtualization

Graphical User Interface (GUI)

Full support for TCP/IP protocol

Open source

GreenCloud is suitable mostly for calculating energy consumption when designing cloud

computing structures. About 80 percent of GreenCloud code is implemented in C++, and the remaining

20 percent is in the form of Tool Command Language (TCL) scripts. To develop new features to

GreeenCloud user supposed to know both of mentioned languages. (“Greencloud - The green cloud

simulator”, 2016).

The architecture of GreenCloud is represented on Fig. 2 .

See Full Size >

Cloud Analyst has been developed at the University of Melbourne with the purpose of evaluating

the the performance of large-scale distributed cloud applications with high user workload that are

geographically distributed over several data centers. Cloud Analyst has been built as an extension of

Features of CloudAnalyst:

Intuitive Graphical User Interface (GUI)

High degree of configurability and flexibility

Repeat experiments with slight modifications

Results are represented also graphically in the form of charts and tables

Ease of extension – permits extending its capabilities to improve the simulation

The architecture of Cloud Analyst is represented on Fig. 3 .

See Full Size >

2.4 EMUSIM

EMUSIM has been developed at the Cloud Computing and Distributed Systems (CLOUDS)

Laboratory, University of Melbourne. EMUSIM provides both capabilities of an emulator and as well of

a simulator of a cloud environment. The simulator is built up on CloudSim and Automated Emulation

Framework (AEF). On EMUSIM, the simulation model is generated with extracted information from

application behaviour via emulation. (“EMUSIM“,2016)

With EMUSIM, cloud service providers, are able now to know now how many resources are

required to have a given response time considering a specific arrival time and the effects of the changes

on the mentioned arrival rate. (Al-Sakib K. P.& Muhammad M. M. & Shafiullah K.,2015)

Features of EMUSIM:

Includes all features of CloudSim

Improved simulation by combining emulation and simulation of cloud computing

EMUSIM does not include a GUI. Cluster configuration is done by editing cluster.xml file. Files

aef.properties and emusim.properties are used to define information about the specific emulation to be

performed

Permits extending its capabilities to improve the simulation by working with CloudSim modules.

See Full Size >

2.5 GroudSim

GroudSim (Gr-Grid and oud-Cloud Simulator) is an event simulation platform for both cloud and

grid computing environments. Main programming language for GroudSim is Java. (Kecskemeti, G &

Ostermann, S & Prodan, R., 2014) (”What are cloud simulators and it's applications”, 2016)

GroudSim features:

It can be extended easily by using probability distribution packages.

GroundEntity module from GroudSim has its own definition for error behaviors. With this

module user can change the configuration during each error occurrence.

The simulation parameters are entered through configuration files in XML format. The

application has no GUI interface

Introduces background load functionality

See Full Size >

2.6 DCSim

DCSim (Data Center Simulator) simulates a virtualized data center providing IaaS service for the

cloud. DCSim is an event-driven simulator designed for transactional and continuous workloads such as a web server. The simulator is developed in Java. The main component of DCSim is the DataCenter, which

contains hosts, VMs, and different management components and policies. The data center is composed of

interconnected hosts that are managed by a set of management policies. Each host it’s composed of a set

of resource managers that manage local resource allocation, a CPU scheduler to decide when to run VMs,

and a power model that decides how much power is being consumed by the host at any point in time. (Al-

Sakib K. P.& Muhammad M. M. & Shafiullah K.,2015) (Tighe, M. & Keller, D. & Shamy, J. & Bauer, M

& Lutfiyya, H.,2013)

Figure 6 outlines the general architecture of DCSim.

DCSim features:

VM live migration and replication

Simulation tasks can be load balanced between multiple application instances running on

different VMs

DCSim generates a log containing the simulation results. This log can be used to generate

graphs, statistics in order to understand the behavior of simulated data center. Simulator includes also a

visualization tool that generates the necessary graphs using as input the generated log file.

Sharing of workload between multiple VM

Introduces SLA Violation parameter to monitor when a VM requires more resources than are

available to it. The percentage of CPU resource required and not available is recorded.

Calculates power consumption for each host

See Full Size >

Conclusion

Each cloud simulator presented covers a certain domain of interest. Unfortunately, on the market

there is no cloud simulator to cover all possible cloud architectures. The user looking to simulate a

specific feature that cannot be found on any cloud simulator can develop his own module with the help of

a certain programming language.

CloudSim seems to be the most used cloud simulator to develop new features. CloudAnalyst and

EMUSIM simulator took advantage of CloudSim toolkit to develop their own cloud computing features.

Cloud Analyst analyses the performance of large-scale distributed cloud applications with high user

workload that are geographically distributed over multiple data centers. EMUSIM combines the AEF

emulator with CloudSim to optimize the simulation results obtained with CloudSim. Java is the most used

programming language among cloud simulators. One benefit of using java programming is portability

between operating systems or even perform mobile simulations (OpenMobster).

Analysing the synthetized results of table 1 we can conclude that CloudSim is the most versatile

cloud simulator solution.

Future work will concentrate on extending the research to other cloud simulators and to structure

them by the ability of extending their features beyond their standard features. Programming language and

ease of adding new features will be taken in consideration.

References

1. Al-Sakib K. P.& Muhammad M. M. & Shafiullah K..(2015). Simulation Technologies in Networking and
2. Communications EMUSIM,DCSIM p338,340
3. CloudSim Simulation Framework - Superwits Academy (2016, August 1). Retrieved from
4. http://www.superwits.com/library/cloudsim-simulation-framework
5. Umang, S. & Ayushi ,S. (2016). CloudSim Simulator Used for Load balancing in Cloud Computing -
6. International Journal of Emerging Technology and Advanced Engineering. Volume 6 Issue 4
7. Greencloud - The green cloud simulator (2016, August 3). Retrieved from
8. https://greencloud.gforge.uni.lu/index.html
9. Boru, D. & Kliazovich, D. & Granelli, F. & Bouvry, P. & Zomaya, A.Y. - Energy-efficient data replication in cloud computing datacenters, Springer Cluster Computing, vol. 18, no. 1, pp. 385-402, 2015. (2016, September 1). Retrieved from https://greencloud.gforge.uni.lu/ftp/energyrep.pdf.
10. EMUSIM: Integrated Emulation And Simulation For Evaluation Of Cloud Computing Applications (2016, August 5). Retrieved from http://www.cloudbus.org/cloudsim/emusim/
11. What are cloud simulators and it's applications - (2016, August 23). Retrieved from https://codingsec.net/2016/08/cloud-simulators-applications/
12. Tighe, M. & Keller, D. & Shamy, J. & Bauer, M & Lutfiyya, H. (2013)- Towards an Improved Data Centre Simulation with DCSim, Proceedings of the 9th International Conference on Network and Service Management (CNSM 2013), P364 – 372
13. Kecskemeti, G & Ostermann, S & Prodan, R. (2014) - Fostering Energy-Awareness in Simulations Behind Scientific Workflow Management Systems, 2014 IEEE/ACM 7th International Conference on Utility and Cloud Computing, P29 - 38