FRAMEWORK FOR PERFORMANCE EVALUATION OF GREEN MOBILE CLOUD COMPUTING USING A CASE STUDY OF IAAS CLOUD DATACENTER.
Introduction
This is a summary of the research that was conducted for my university undergraduate final year project.
Outline
1. Summary
2. Motivation
3. Contribution to knowledge
4. Solutions/Methodology
5. Research Implementation
6. Results and discussion
7. Solutions to implement green cloud computing with techniques from this research
8. Conclusion and recommendation
1. Summary
The research conducted for my university final year project was based on a framework for performance evaluation of green mobile cloud computing. I used a case study of a cloud infrastructure as a service (IAAS) which I simulated with Apache Cloudstack and Cloudsim, Cloudsim was used as a simulation framework for mobile cloud computing. Cloudstack was used to build the IAAS cloud datacenter, which I then did performance evaluation on the Ubuntu virtual machine running this Cloudstack using Phoronix test suite (PTS) . Finally, I performed Power usage effectiveness(PUE) and data center infrastructure efficiency (DCIE) calculations from the benchmark data, which I used as a solution for green mobile cloud computing.
2. Motivation
The widespread deployment of IT has had inadvertent side effects, such as increased energy consumption and pollution. Technology-related power consumption is rising rapidly.
According Julie Zaugg article in CNN ; China’s data centers emit as much carbon as 21 million cars.
Green mobile cloud computing is needed to help cloud providers implement adequate minimization of power consumption for energy efficiency. It also helps to minimize and eradicate the release of harmful carbon gases to the environment by datacenters. Therefore there is a need for performance evaluation of green mobile cloud computing using a suitable framework.
3. Contribution to knowledge
At the end of this research a Cloudstack and Cloudsim simulation framework for performance evaluation of green mobile cloud computing was provided.
4. Solutions/Methodology
We used two of the techniques for green mobile cloud computing which are;
(i)Monitoring and metering by PUE and DCE
(ii)Virtualization of servers with Apache Cloudstack.
5. Research Implementation
5.1 The first step was the system Simulation
The simulation detail encompass the operation, the performance and modelling of the system.
5.2 System simulation of Cloudsim Framework for mobile cloud computing.
Running the simulation with cloudsim on eclipse IDE to simulate how mobile cloud computing is implemented
5.3 The steps for simulation of Cloudstack framework
The framework for performance evaluation of the IAAS cloud datacenter was built using cloudstack running on the virtualized Ubuntu 18.04 system in virtualbox hypervisor. The Latest version (cloudstack 4.15) was used for the simulation.
5.3.1 Add the Ubuntu VM to Oracle Virtual box hypervisor
5.3.2 Power up Ubuntu vm
5.3.3 Cloudstack running on Ubuntu 18.0 VM
5.4 Steps for Performance evaluation and benchmarking using Phoronix test suite (PTS) tool on the Ubuntu 18.04 vm.
First install the PTS tool on Ubuntu 18.04 virtual machine (vm) using the command “sudo apt-get install -y phoronix-test-suite” , then run the following commands shown in figures below to perform the performance evaluation and benchmarking.
6. Results and discussion
This simulation involved creating a data center with one host and runing one cloudlet on it. The results show that the cloudlet was created in 400 seconds. This simulation demonstrated how much time it took for a mobile to connect to a cloudlet, that runs in a datacenter.
A score of 564/ 600 MIPS: this means that processor performance of the Ubuntu 18.04 VM is performing well and above average
A score of 178/200 TPS: this means that Disk I/O performance of the Ubuntu 18.04 vm is performing well and above average.
A score of 5447.54/6000 mb/s: this means that the memory performance of the Ubuntu 18.04 vm is performing well and above average.
A score of 4904/5500 Mbits/Sec: this means that the network performance for TCP of the Ubuntu 18.04 VM is performing well and above average
Therefore a score of 1.05 /1.1815 Mbits/Sec: this means that the network performance for UDP of the Ubuntu 18.04 VM is performing well and above average.
The launching of the cloudstack and running of the virtual machines on cloudstack , increases the power used by the Ubuntu machine.
7. Solutions to implement green cloud computing with techniques from this research
The key green cloud computing techniques or solutions applied in this research are :
I. Virtualization
II. Monitoring/Metering
7.1 Energy efficiency for green mobile cloud computing through virtualization.
Effects of virtualization on power consumption
Virtualization helps to make data centers green and enhances energy efficient so as to ensure that IT infrastructure contributes as little as possible to the emission of green house gases through server and resource utillization, live migration, data deduplications and data shrinkage. Virtualization provided better utilization of resources available in the Iaas cloud data center, thereby minimizing power usage.
7.1.1 Virtualized System used in this research.
Virtualized System used in this research is divided into three components
7.2 Energy efficiency for green mobile cloud computing through monitoring and metering
The standard calculations used to achieve energy efficiency in a cloud datacenter are PUE and DCIE ,which are used for monitoring and metering.
To measure the unified efficiency of a datacenter and improve its’ performance per-watt, the Green Grid has proposed two specific metrics known as the Power Usage Effectiveness (PUE) and Datacenter Infrastructure Efficiency (DciE).
(i) Power Usage Effectiveness metric (PUE) :The PUE is intended to help operators understand a data centre’s efficiency and reduce energy consumption. It is defined as the ratio of total data centre input power to the power used by the IT equipment.
PUE = Total Facility Power / IT Equipment Power
The Total Facility Power is defined as the power measured at the utility meter that is dedicated solely to the datacenter power. Datacenter facility units include: Pumps, Cooling tower, Generators, chillers, battery, Uninterruptible power supplies (UPS)
(ii) Datacenter Infrastructure Efficiency (DciE) : Data Center Infrastructure Efficiency (DCiE) is a metric used to determine the energy efficiency of a data center. DCiE was developed by members of the Green Grid, an industry group focused on data center energy efficiency. DCiE, which is expressed as a percentage, is calculated by dividing IT equipment power by total facility power. (Biswajit, 2018).
DCiE =1 / IT Equipment Power/Total Facility Power x 100%
DciE = 1/PUE x 100%
7.2.1 Using power consumption benchmark data to calculate PUE and DCE
To calculate the exact efficiency value, the values are required namely total power consumed by facility and total power consumed by IT equipment.
(I) Total power consumed by facility: Facility was not used in the iaas cloud datacenter in this research. So we used a watt meter with the computer running the Ubuntu vm plugged into it.
(II) Total power consumed by IT equipment :IT equipment used in the iaas cloud datacenter in this research are: A computer which serves as a server in the datacenter. results were obtained by monitoring the computers power system sensors.Results were obtained and analyzed.
Total IT Equipment Power = 4.5W + 0.9W + 0.4W + 1.7W = 7.5W
Total Facility Power = 15 W (based on using a watt meter with the computer plugged into it)
PUE = Total Facility Power/Total IT Equipment Power PUE = 15W/7.5W = 2W
PUE = 2W
DCIE = 1/2 X 100= 50%
The results also show that the overall performance of the Iaas cloud data center in energy efficiency was averagely efficient with a PUE value of 2watts based on using a watt meter with the computer plugged into it which generated a Total Facility Power of 15 for the Iaas cloud datacenter. A 50% DCIE value indicates average.
Effects of monitoring and metering on power consumption
The results demonstrate that PUE is a standard method to be used in measuring the performance of the data center. PUE metrics help to show that managers are aware of green and energy-efficient concepts.
8. Conclusion and Recommendation
8.1 Conclusion
The research conducted has been able to ascertain that :
The virtual machine running cloudstack performed well and above average in terms of CPU performance, disk I/O performance, Memory performance, Network performance and archieved scores more than average needed for a correctly functioning system
The research also revealed that the green mobile cloud computing based on the framework for performance evaluation of Ubuntu VM operation shows that:
Energy efficiency and power minimization can be achieved by Virtualization which prompts green cloud computing through server utilization, and monitoring and benchmarking power consumption metrics can help determine the datacenter DCE and PUE
8.2 Recommendation
The research is recommended to reduce power consumption in a cloud IAAS datacenter by virtualization, and monitoring and metering, thereby promoting green mobile cloud computing.
8.3 Future Work
The future work may entail the use of more than one virtual machine to setup on CloudStack management server , so we have 2 host or more for performance evaluation. Also performance evaluation and benchmarking can be done on a physical real life datacenter instead of simulation.
WRITER: OLUYEDE SEGUN . A(jr)
twitter profile: https://twitter.com/oluyedejun1
linkedin profile: https://www.linkedin.com/in/oluyede-segun-adedeji-jr-a5550b167/
TAGS: #cloudcomputing #ApacheCloudstack #Cloudsim #greencomputing #mobilecomputing #Ubuntu
