Minimizing Power Consumption and Improve the Quality of Service in the Data Center

Minimizing power consumption and improve the quality of service in the data


Introduction
The computing environment consumption of energy is the main issue in the IT industry, environment and academic. Because of energy usage in data center for its operations such as storage and recover, supply of power and cooling contributed completely on total cost for operations 1 . Cloud framework 2 aims to advanced design, through image processing, virtual machine management and scheduling of VM. The power usage in the data center has become the main issue for cloud computing. Storage and power usage in the data center are mounting in day to day life. Now a days, Amazon, EC2, IBM, Microsoft and Google and many big companies have deployed data centers for hosting Internet applications. In this paper a framework of cloud computing which is directly contributed towards the energy consumption in the data center is discussed, and also how green cloud computing lowers the power consumption has been discussed. The deployed framework is fully dependent on open stack net with 10 different places tested by multiple physical nodes. Quality of Service (QoS) presented by the application of availability, reliability, and performance of different levels with infrastructure and platform hosts it. Framework and QoS are fundamental for the cloud environment, who is delivering the excellencet characteristics and tradeoffs QoS different levels of operational cost.
The data center cost and energy consumption driven in the cloud environment can be reduced as much as possible through VM selection on the workload basis 3 .
Energy efficiency in the cloud environment and increase in the efficiency through VM in the data center will provide Return on Investment (ROI) to the environment and reduce CO 2 emission which is directly dependent on the globe 4 .
The surveys made in the energy efficiency of data center have found different techniques, they are 5 : • Reducing electricity bills • Minimizing energy with interconnected networks • Renewable energy. According to Google, Microsoft, IBM, EC2 and Amazon data centers, the energy in IT industry is nearly 95% of the total consumption in the year 2015 in that the servers used 80%, switches 8% and other facilities 13% of energy.

Framework
The deployment of open stack is fully depending on open source software which is dynamic consolidation by VM. The open stack is designed for VM consolidation dynamically through the environment as Infrastructure as a Service (IaaS) in the cloud environment. The data centre consisting of N different physical nodes. Each and every node i is characterized by the Million Instructions Per Second (MIPS). Amount of power usage per second by the server and different network components is huge.
To store information, server not having direct relationship between network equipments and storage networks to allocate VM. The environment type implies network and workload configuration with time on the provisioning of VM. Multiple users request multiple characteristics to M heterogeneous VMs to process power in MIPS and depend on the network bandwidth and how much utilization of Random Access Memory (RAM) is utilized in the server.
The individual users are requested by the super admin then super admin will check the Local admin to look over the availability of space in the server and convey to the super admin. Again super admin is requested by Virtual Machine Manager (VMM) to allocate the provisioning of workload and power utilization by the server, then each VM to migrate individually through physical node and N physical nodes 7 . The users will get establishment of the QoS improvement in the data center.
In this paper it is discussed about the selection of VM depending upon the workload condition by keeping time factor as a parameter. If the host is overloaded the local admin behind the overloaded host initiates the VM selection process to identify the VM offload working in the host 5 .

Power Management in Data Center
Virtualization is the main key in the continuous improvement of power management in cloud computing. the servers and costs are the two factors considered in the energy efficient cloud environment. The fact that closely (80 to 82) % of workload in server is supported by X86 hardware component are in virtual machine in the year 2018.
Usage of virtualization is more popular and very easy to understand. The benefits among virtualization in the data center are: • Hardware expenses: Used large number of lower capacity servers. Instead of using many physical hosts machines it is replaced into virtual ones. It consolidates the total power consumption used is significantly lower. • Increase availability: During server utilization with "VM allocation" the Capabilities of VMs relocate and the feature of physical host connectivity quickly in the business field keep track of applications up and running simultaneously 8 .

Proposed Mechanism
There are many ways to define the usage of servers and improve the QoS in the data center. For example: if any of the application run on the servers power utilization 8 to 10 years old is the twice the power utilization in the newer servers and new servers are most efficient servers 9 . A simple advanced relationship between Central processing Unit (CPU) utilization and power utilization is engaged with power model in the data center of the cloud environment. VM allocation and selection process is enhanced through the memory usage in the data center and even it requires some extra memory to be considered as a buffer temporarily. The VM migration policy is proposed towards the time and CPU utilization during the time period is considered as 10% of energy is consumed during migration policy 10 . The VM migration is accurate and depends on the time period in selecting the bandwidth. The Service Level Agreement (SLA) is placed by considering the whole CPU utilization and all the networking components. The Million Instructions per Second (MIPS) can be obtained purely depending on the total number of hosts who requested the work on demand to VM in the data center.

VM Selection Process
VM allocation through the server will reduce the energy consumption in the data center. The VM allocation and process through the selection can be described as the following process.
VM selection process begins with the number of physical hosts with time as a main factor and CPU utilization with RAM is considered. Initially the minimum workload allocated VM and RAM are to be considered. The maximum CPU utilization with different workload condition to be identified. Select the VM which is empty at initial condition. Consider the VM and CPU utilization with maximum and select the VM RAM which is greater than the RAM and continue till the completion of the process then calculate mean while selecting sum of the n from CPU and length 6,11 .
The VM selection process includes P as a parameter represents the total VM performance of object j. C is the co-efficient of the average performance used in VMs. Whereas CPU utilization in the cloud as 0.2(20%) of the approximation.
For test the environment of data set considered R1, R2,….Rn is used to calculate the utilization of memory and mean of the total VMs considered 7,12 . P=C.ᶋt0+Tnj qj(t)dt---(1) Rnj=Mj/Bj ---(2) Whereas qj(t) is the time utilization of Vmj, t0 shows the time to process the VM to physical host. Rnj is the completion of the process. Mj used by utilization of memory, Bj is the bandwidth usability.

Experimental Setup and Results
In the cloud environment the Infrastructure as a Service (IaaS) is considered with infinite resources to the users. Cloud improves the Quality of Service through availability, reliability and performance applicable in the (IaaS) Level 13 . QoS improve the cloud service and emission of Carbon Dioxide CO 2 . The software environment Cloud Sim tool kit is a modern framework which is used to test many simulation experiments. The energy consumption in the various storage devices is up to 70 to 80% of the total utilization of hardware components 14,15 . The experimental data centre consists of 200 physical nodes. Among the many are HP LiantG1 servers and rest are used with HP LiantG2 servers. The power consumption in VM consolidation data centre is given in the table 2. The servers G1 and G2 are having 1GB/S network bandwidth.
HP LiantG1 (Intel Xeon 1020, 2 cores X 1240 MHZ 2GB) and HP LiantG2 (Intel Xeon 1025, 2 cores X 1220MHZ, 2GB) the power consumption in the Table 2 shows the power consumption with different servers in watts. VM selection process performs an efficient result with VM consolidation for every 10% difference from 0 to 100% has verified. The simulating results in CPU is advantageous by overloading a server a lighter workload is required. VM consolidation shows the result by considering the different number of VMs assigned with workload with different days and calculated mean by selecting total number of physical nodes and workload in different servers used by the day with the number of servers actively working on that day. The graph represents the total number VMs considered in the particular date with different workload conditions and calculated mean to be shown in VM consolidation and improved Quality of Service.  The graph represents power consumption with different servers in watts with VM consolidation for every 10% differences from 0 to 100% as plotted.

Conclusion
The experiment compares the power consumption in the data center through VM selection process. The main aim is to reduce the power consumption in data center through VM selection process. Power consumption in different workload conditions is shown in Figure 5. The experimental results show an efficient result with VM consolidation for every 10% differences from 0 to 100% has verified. The simulating results of CPU are advantageous by overloading to a server. The different number of VMs assigned with different workload condition in the data center and the energy is reduced by 35% on the total impact and Qos is improved. Future work can be carried out to reduce the VM selection process and increase VM performance in the data center with large number of physical nodes in the data center.