Improved Power Efficient VM Selection
Strategy to Consolidate Workload in
Compute Cloud
Heena Patel
PG student1Dr. Hiren Patel,
Professor2Prof. Nimisha Patel
Associate Professor1Research Scholar3
1 Sankalchand.Patel College of Engineering Visnagar-Gujarat, India 2 LDRP Institute of Technology & Research, Gandhinagar-Gujarat, India
3 Rai University, Ahmendabad-Gujarat, India
Abstract :
Computing is a use for processing power, storage, memory and other computational resources are shared or rented to customers over Internet. The services are offered through virtualization technology in which resources in form of virtual machines are allotted to users. In Cloud datacenter, virtual machines are hosted on physical machine. Huge amount of power consumption by Cloud datacenter impacts significantly on environmental and economical aspects. In this research work, we aim to reduce energy consumption in along with boost the utilization of resources. In this paper, we propose an improved power efficient mechanism which efficiently makes use of VM selection strategy and consolidate workload. We wish to achieve minimization in resource requirement while not compromising the Quality of Service parameters.
.Keywords: Cloud computing, VM Selection, Load Balancing, Workload Consolidation.
1. Inroduction
Cloud Computing is widely adopted by business, individual and large enterprise. Cloud Computing is user demand network access for sharing pool of resource like network, storage, application and services. This resource can be rapidly provisioned and released with minimal management effort or service provider interaction [1]. The multiple data centers which is supported by many network services and the data centers require mass of energy, high operating costs in a result [2]. In recent year, most researchers have try to reduce energy consumption and make that is maximum useful and environmental beneficial.
Virtualization technology is a backbone for working of Cloud environment. Implementation of virtualized resources involves specialized hardware, software, or a combination. The virtual machines is independent it does not require specialized and hypervisor-specific hardware. Virtualization does require more bandwidth, storage and processing capacity than server or desktop if the host multiple virtual machines running on host. Virtualization technology very special for improving computing resource utilization of in data centers by creating multiple virtual machines on a single physical server or host and implementing the VM consolidation. VM consolidation is a one technique to consider the aggregated resource consumption of particular VMs, to avoid reduces performance and Service Level Agreement (SLA) violations. VM consolidation classifies the workload on different virtual resources situated on different physical server. Each VM requires a certain amount of resources such as CPU, memory, bandwidth, storage etc. Virtualization provides resource based on user demand. VM migrations require extra management cost for example virtual machine composition, online VM migration, creates a VM, which causes extra energy consumption. so, we try to reduce the number of VM migration to reduce energy consumption. This work aims to work on defining a VM selection policy with objectives of workload consolidation, reduction in migration in SLA violation.
2. Related Work:
The Cloud Computing is maximum used by ever one, for attracting user cloud providers are paying to the cost and efficiency of the data centers. Cloud providers provide Quality of services at lowest cost. So many researchers aim to reduce energy consumption. Beloglazov et al. [3] and [4] have found that VM consolidation is carried out by four scenarios viz. (1) Underloaded nodes to put in sleeping mode and migrate all VMs on active hosts (Underload detection). (2) Some VMs are migrated from overloaded host to other host for QoS preservation (overload detection). (3)VMs should be selected for migration on appropriate target host (VM selection). (4) Placing selected VM on active host (VM placement).
Filipe et al. [5] have proposed a new VM selection policy which is known as VBalance, and it is used for reducing energy consumption and to maintain SLA between service provider and user in Cloud data center. The VBalance policy takes different input parameters. The Coefficient which is Minimum Determination has some five values, which significant that VBalance policy will work on function of MLR (Multiple Linear Regression) is equal or greater than the number of migrations at VMs for great energy economy.
Fu et al. [6] define VM selection policy used for reduce the energy consumption without affect on SLA violation in virtualization host. They propose VM selection policy called Meet Performance which considers degree of resource satisfaction and can reduce energy consumption, VM migration time and SLA violation. Ahmed et al. [7] Propose a VMC (Virtual Machine Consolidation) framework based on host categorize. The common VMC process is to first migrate VMs from overloaded host then migrate VM form under loaded host. The VMC problem is dived into four sub-problems: overloaded host, under loaded host, select VMs to migrate and place the selected VMs. Here authors use VM selection policy which is Maximum Requested Bandwidth (MBW). MBW for select VM migrate from overloaded host that has Maximum requested bandwidth. Alaul et al. [8] redesign VM selection algorithms combining migration control and heuristics method. The main objective of this research is to reduce power consumption. After simulation they found that the new access methods and reduces power consumption.
In CloudSim [9], server is checked status at every interval time when overloaded host are identified VM, which is select for migration. However there are places where modification can be done and get expected QoS. We propose to implement the work in CloudSim as part of our future work.
Comparing the previous work, hosts in Cloud data center are divided in three categories either overloaded, under loaded and normal. In this work, we propose a modified VM selection algorithm which is used to select the VMs to be migrated from overloaded hosts in such a way that the targeted overloaded host becomes normal. Our VM selection method aim to reduce energy consumption and meeting the QoS requirement.
3. Our Proposal
Since multiple VMs are run on a single host, if the host is overloaded, we need to select some VMs for migration. If random selection policy is applied in VM consolidation, some VMs that operate efficiently may be moved while those that use many resources and provide poor efficiency still run normally. It not only increases energy consumption of data center but also reduces the utilization of resources. Therefore an efficient VM selection policy is required in VM consolidation.
There are three basic VM selection methods (i) Minimum Migration Time (MMT) (ii) Maximum Correlation (MC) (iii) Random Selection (RS). Here, we aim to add some parameters in modified VM selection method. Our aim is to reduce power consumption and reducing migration. So we use VM selection strategy to consolidate workloads in compute Cloud.
VM Selection Algorithm:
Input: overloadedHost
Output: listVMsToBeMigrated Weight to standard deviation α Weight to average utilization β
1. for each i varying from 1 to vm.size() do if(α > 20) then
//vm is High fluctuate
list VMsToBeMigrated.add(vm(i)) increment i
endif
2. endfor
3. // sort the vm based on the standard deviation and average utilization 4. for each i varying from 1 to vm.size()-1 do
for each j varying from i+1 to vm.size() do
rank-i <- vm(i).getStandardDeviation() * α + vm(i).getAverageUtilization() * β rank-j <- vm(j).getStandardDeviation() * α + vm(j).getAverageUtilization() * β if(rank-i > rank-j) then
tempvm <- vm(i) vm(i) <- vm(j) vm(j) <- tempvm endif
endfor
5. endfor
6. isOverloaded <- TRUE
7. utilization <- overloadedHost.getUtilization() 8. while isOverloaded is TRUE do
utilization <- utilization - vm(i).getUtilization list VMsToBeMigrated.add(vm(i))
if utilization < upperThreshold then isOverloaded <- FALSE
else increment i endif
9. endwhile
10. return listVMsToBeMigrated
Here, in the algorithm of VM Selection, first we take the assign weight to standard deviation (α) and average utilization (β). These weights (α and β) are used to calculate rank of each VM. We sort VMs based on the rank. Further, for each host, we compute the resultant utilization after deducting VM one by one until the targeted host becomes normal. The list of the targeted VM is returned. The flowchart of the algorithm illustrated in Figure 1.
Flowchart:
Fig 1: Flowchart of VM selection method
The proposed algorithm is only theoretically tested. In next phase of our work, we aim to implement our VM selection policy on CloudSim [9] simulator and test it with real dataset provided by PlanetLab [10]. The outcome of experimentation will be compared with that of contemporary policies.
4. Conclusion
The issue of energy consumption in Cloud datacenter has attracted the attention of research community. In recent year, researchers try to reduce energy consumption and SLA violation. In this work, we propose a new method for selecting a VM from overloaded host. Selected VM is migrated from overloaded host to other host. In our paper, we use a VM history data to categories VM in three types VM viz. (1) Steadily with upper utilization (2) Steadily with lower utilization and (3) Highly Fluctuate. We added new schema in modified VM selection algorithm. Our aim is to reducing energy consumption, reducing in VM Migration and also reducing in SLA violation.
[2] Zhu X, Young D, Watson B J, Wang Z, Rolia J, Singhal S, McKee B, Hyser C, Gmach D, Gardner R, Christian T, Cherkasova L. 1000 islands: an integrated approach to resource management for virtualized data centers. Cluster Computing, 2009, 12(1): 45–57
[3] A. Beloglazov, J. Abawajy, and R. Buyya, “Energy-aware resource allocation heuristics for efficient management of data centers for Cloud computing,” Future Generation Computer Systems (FGCS), vol. 28, no. 5, pp. 755–768, 2011.
[4] A. Beloglazov and R. Buyya, “Optimal online deterministic algorithms and adaptive heuristics for energy and performance efficient dynamic consolidation of virtual machines in Cloud data centers,” Concurrency and Computation: Practice and Experience (CCPE), vol. 24, no. 13, pp. 1397–1420, 2012.
[5] Filipe Fernandes S B de Matos, Joaquim Celestino , André RibeiroCardoso “VBalance:A Selection Policy of Virtual Machine for Load Balancing in Cloud Computing” 978-1-4673-7194-0/15/$31.00 ©2015 IEEE
[6] Xiong FU , Chen ZHO“Virtual machine selection and placement for dynamic consolidation in Cloud computing environment” Springer– 2015
[7] Dabiah Ahmed Alboaneen1 2, Bernardi Pranggono1 and Huaglory Tianfield1” Energy-aware Virtual Machine Consolidation for Cloud Data Centers” 978-1-4799-7881-6/14 $31.00 © 2014 IEEE DOI
[8] Mohammad Alaul Haque MonilLee, Rashedur M Rahman,” Implementation of Modified Overload Detection Technique with VM Selection Strategies Based on Heuristics and Migration Control”978-1-4799-8679-8/15/$31.00 copyright 2015 IEEE.
[9] R. N. Calheiros, R. Ranjan, A. Beloglazov, C. A. F. D. Rose, and R. Buyya, “CloudSim: A toolkit for modeling and simulation of Cloud computing environments and evaluation of resource provisioningalgorithms,” Software: Practice and Experience, vol. 41, no. 1, pp. 23–50, 2011.