A Systematic Review of Existing VM Migration Techniques

Ramandeep Kaur, Supreet Kaur
2017 International Journal of Advanced Research in Computer Science and Software Engineering  
The virtual machine migration is core feature of virtualization that plays important role in cloud computing. The resource utilization is monitored by local migration agent that launches migration of virtual machine from one physical machine to another. Various virtual machine migration techniques are explored for efficient utilization of resources. With rapid increase in data centers, it becomes important to design virtual machine migration. Hence, in this paper we review, various techniques
more » ... r virtual machine migration with their merits and demerits. IV. REVIEW OF LITERATURE Wen-tao Wen[2015][1] proposed ant colony optimization algorithm using live migration .The algorithm described that local migration agents continuosly monitor resource utilization and then launches migration of virtual machines. Ramezani, FahiMeh [2014][2] proposed task based scheduling algorithm with particle swarm optimization techniques. The algorithm transferred extra tasks from overloaded virtual machine instead of transferring the entire virtual machine. Gutierrez-Garcia, J.Octavio [2015][3], proposed agent based algorithm capable of balancing load across commodity. As a result, number of migration are reduced. Zhu, chang peng,Bo Han[2015][4]proposed queuing theory and bandwidth allocation algorithm. The queuing theory modeled relationship between virtual machines. The algoritm improved effect of live virtual machine performance. Qiang Li[2009][5], author utilized feedback algorithm to manage virtualized resources and machines together into shared pool. Following SLO agreement the VM allocation takes place. Bo Li, Jianxin Li[2009][6] proposed algorithm based on distribution of workload in virtual machines with workload migration. The workload resize virtual machine migration. Babu, KR Remesh, Amaya Anna[2015][7] proposd Bee colony algorithm that considered the task priorities of virtual machines waiting in queues.The algorithm improved quality of services(QOS). Hu, Jinhua, Jianhua Gu, Guofei Sun[2010][8] proposed genetic algorithm that considered historical and current state of the system. The algorithm computed influence on host after VM resources are deployed on it. It then chose least affected solution. Gregor von Laszewski ,LizheWang[2009][9] proposed energy efficient algorithm that scheduled virtual machines in a compute cluster so as to reduce consumption of power consumption of power through DVFS. Rajkuma Buyya[2015][10] proposed a model that captured user profiles and supported simulation of resources utilization in cloud environment. The model validated that there is a need of different statistical distribution to represent CPU, memory and total number of instructions. Keng-Mao Cho[2014][11] proposed hybrid meta-heuristic algorithm that combined ant colony optimization and particle swarm optimization. The algorithm added a PSO operator into ACO algorithm to reduce computing time and improve virtual machine migration. Chen, Kun-Ting, Chien Chen[2014][12] proposed network aware migration algorithm that obtained compatible multi resource LB performance. Komarasamy, Dinesh[2016][13] proposed effective bin packaging algorithm in which virtual machine are dynamically split and joined based on speed of proposed resource. Jayasree, P., and Meenakshi Sharma[2014][14] proposed cloud booster optimized virtual machine algorithm in which VM resource allocation was covered out at both server weights and for future predictions. Sarker,Tusher Kumer [2009][15] proposed heuristic algorithm and deadlock handling a algorithm that migrated multiple virtual machines on data center at same time with minimum migration time and downtime . Jinhua Hu[2010][16] presented a scheduling strategy on load balancing of VM resources based on genetic algorithm. This strategy solved problem of load imbalance and high migration cost by traditional algorithm after scheduling. Anton Beloglazov[2010][17] modified Best Fit Decreasing[MBFD] algorithm and presented some evaluation results that showed that dynamic reallocation of Virtual machine bring substantial energy savings. Ching-Chi Lin[2011][18] proposed dynamic Round robin algorithm that reduced significant power consumption and consolidated energy-aware virtual machine. V. VM MIGRATION TECHNIQUES The various VM migration techniques that enable live migration are: 1) Ant colony optimization VM migration(ACO-VMM) : ACO-VMM[1] is ant colony optimization algorithm in which the resource utilization is examined by local migration agent in an independent manner and then launches the migration. The main goal of ACO is to search minimum cost graph in the graph. The current and last condition are taken into account to avoid unnecessary migrations. Wen Tao Wei, proposed ACO optimization algorithm[1] which was based on Ant behavior searching for food so that the real optimal solution to be found. In this algorithm like the ants the virtual machine will leave more Pheromones(Markers) when the bandwidth between the original and destination physical machine is less or when the destination physical machine has a higher load condition. Results: The number of migration were reduced by using ACO-VMM. It control the load balance variance and improved SLA violations. 2) Ant Colony System-based VM Consolidation(ACS-VMC) ACS is an optimization metaheuristic algorithm for migrating the live virtual machines so that the under loaded physical machines can be powered off or put into low power state. Fahimeh Farahnakian[], proposed ACS-VMC approach that used artificial ants for consolidating virtual machines into less number of active physical machines based on their current resource requirements. It outperformed existing VM consolidation by maintaining the desired QOS. The physical machine status (overloaded, under-loaded, normal, predicted-overloaded) was detected by local agent. Results: This algorithm reduced the consumption of energy and avoided SLA violation. It also calculate the number of VM migration.
doi:10.23956/ijarcsse/v7i5/0180 fatcat:nfv5d34vkrb2vdvflc2yjf7syy