Comparative Study of Virtual Machine Placement Algorithms in Cloud Computing Environment
VIRTUAL MACHINE PLACEMENT PROBLEM
CLASSIFICATION OF VM PLACEMENT ALGORITHMS
A. Constraint Programming
In this technique some constraints are applied and these constraints must be fulfilled on relations between variables. The VM placement problem can be designed as a constraint programming problem. These constraints could be Capacity, SLA, QoS.
B. Bin Packing
The VM placement problem can be designed as a bin packing problem. The PMs can be considered as bins and the VMs to be placed can be considered as objects to be filled in the bins. The aim is to place as many VMs into a single PM so that number of PM required to pack the VMs is minimized.
C. Stochastic Integer Programming
Stochastic Integer Programming is used to optimize the problems, which involves uncertainty. VM placement problem can be considered as a Stochastic Integer Programming because resource demand of the VM are known or it can be estimated and the objective is to find the suitable host which consume less energy and minimize the resources wastage.
D. Genetic Algorithm
A GA is used to find exact or approximate solutions to optimization and search problems. The VM placement problem can be designed as a genetic programming problem.The solution domain can be represented as the PM with a resource provisioning capacity. The fitness function can be defined over the number of bins in the solution. The aim would be to deliver a solution that is nearly optimal in terms of the number of bins used and the efficiency of packing of the bins.
E. Simulated Annealing Algorithm
There are certain optimization problems that become unmanageable using combinatoric methods as the number of objects becomes large. A typical example is the traveling salesman problem. While this technique is unlikely to find the optimum solution, it can often find a very good solution, even in the presence of noisy data. The VM placement problem can be designed as a simulated annealing problem. Initially a random solution is generated that could be the mapping of VMs to PMs. Then cost is calculated for this mapping after defining a cost function which could be minimum number of migrations, least power consumption or resource leakage. Then random neighbor solution is generated and its cost is calculated. Then reshuffling of VMs to PMs is done according to the minimum cost.
 A. Beloglazov et al., “Energy efficient allocation of virtual machines in cloud data centers”, proceeding in 10th IEEE/ACM Intl. Symp. on Cluster, Cloud and Grid Computing, pp. 577-578, may 2010.
 Lei Xu,Wenzhi et al., ”Smart-DRS:A Strategy of Dynamic Resource Scheduling in Cloud Data Center”, IEEE International conference on Cluster Computing Workshops, pp. 120-127, Sept..2012.
 N. Bobro et al., “Dynamic placement of virtual machines for managing SLA violations”, proc. in 10th IEEE international symposium on integrated network management pp. 119-128, May 2007.
 Yongqiang Wu, Maolin Tang and Warren Fraser. A Simulated Annealing Algorithm for Energy Efficient Virtual Machine Placement. 2012 IEEE International Conference on Systems, Man, and Cybernetics October 14-17, 2012, COEX, Seoul, Korea.
 Sosinsky, “Cloud Computing Bible”, Wiley Publishing Inc. (2012).
 Kim, “Experimental Study to Improve Resource Utilization and Performance of Cloud Systems based on Grid Middleware” proceeding in first International Conference on Internet (2009).
Please use Google Chrome to obtain the best export results.
Public - 6/27/16, 5:41 AM