CAJ | 학술논문

针对当前虚拟机迁移算法易造成物理节点负载不均衡、能耗高和SLA违反率高等问题，提出了一种基于云计算的在线迁移算法。通过构建在线迁移算法模型和虚拟机负载评估向量，依据各物理节点的带宽占用率、内存占用率和CPU占用率来确定迁移时机和需要迁移的源物理节点；同时，依据各节点的CPU占有率、ROM使用效率和带宽的利用率来确定目标物理节点。通过仿真模拟试验得出，该算法能在虚拟机在线迁移中大幅度提高物理节点的负载均衡，且其SLA违反率和云计算数据中心能耗都最小。
침대당전허의궤천이산법역조성물리절점부재불균형、능모고화SLA위반솔고등문제，제출료일충기우운계산적재선천이산법。통과구건재선천이산법모형화허의궤부재평고향량，의거각물리절점적대관점용솔、내존점용솔화CPU점용솔래학정천이시궤화수요천이적원물리절점；동시，의거각절점적CPU점유솔、ROM사용효솔화대관적이용솔래학정목표물리절점。통과방진모의시험득출，해산법능재허의궤재선천이중대폭도제고물리절점적부재균형，차기SLA위반솔화운계산수거중심능모도최소。
As current virtual machine migration algorithm is likely to cause physical node load imbalance, high energy consumption, SLA violation rate and other issues, this paper proposed a new algorithm based on online migration of cloud computing. By building models and virtual machine load assessment vector, through bandwidth usage based on each physical node, memory usage and CPU utilization, this paper determined migration timing and source physical nodes; meanwhile, according to the CPU usage of each node, the efficient use of ROM, and bandwidth utilization, it determined the target physical node. Finally, the simulation of environmental testing proved that the proposed algorithm in a virtual machine live migration could greatly improve load balancing of the physical nodes, minimize the SLA violation rate and the power consumption in cloud computing data center.