The Virtual Machine(VM) placement is a serious problem to limit the improvement of resource utilization of data center. The VM traffic bandwidth demand is a Non zero-sum resource that the global traffic sum is relativ...The Virtual Machine(VM) placement is a serious problem to limit the improvement of resource utilization of data center. The VM traffic bandwidth demand is a Non zero-sum resource that the global traffic sum is relative with each VM placement position. In this paper, we introduce a new improved traffic constant algorithm in the data center, called Degree and Weighted Maximum Traffic Ratio(DWMTR). The proposal DWMTR algorithm defines a new weighted ratio parameter in this paper. The main body of the parameter is constructed with the ratio, current overall intra-cluster traffic divided by current overall inter-cluster traffic, when a new VM places in the data center. The DWMTR algorithm has the ability to constraint the inter-cluster traffic incensement more strictly than the current VM placement algorithms based on traffic bandwidth allocation. For this algorithm based on the theoretical analysis and simulation, it confirms the proposed DWMTR possesses smaller global interactive traffic cost than the control group algorithms in the appointed VM placement in the three-layer data center model.展开更多
云计算作为一种新的商业计算模型,提供了弹性计算和存储资源等服务.云计算具有超大规模、高可扩展性、高可靠性、虚拟化、按需服务和价格低廉等特点,许多用户把应用提交到云环境下运行.随着越来越多的数据密集型应用部署到云计算环境上...云计算作为一种新的商业计算模型,提供了弹性计算和存储资源等服务.云计算具有超大规模、高可扩展性、高可靠性、虚拟化、按需服务和价格低廉等特点,许多用户把应用提交到云环境下运行.随着越来越多的数据密集型应用部署到云计算环境上运行,应用会有不同的QoS要求.为了保证在放置应用数据副本时,数据副本都能满足应用的QoS要求,我们需要考虑云计算环境下QoS约束的应用数据副本放置问题,为此提出一种QoS感知的副本放置算法(QoS-Aware Replica Placement with PSO,QRPPSO),该算法首先对QoS约束集的属性进行量化,然后构建目标优化模型,最后在副本放置优化算法引入粒子群优化来决定副本放置位置.仿真实验结果表明所提策略能有效解决QoS约束的应用数据副本放置问题.展开更多
基金supported by National Major Projects (No. 2015ZX03001013-002)National Natural Science Foundation of China (No. 61173149)+1 种基金Beijing Higher Education Young Elite Teacher ProjectFundamental Research Funds for the Central Universities
文摘The Virtual Machine(VM) placement is a serious problem to limit the improvement of resource utilization of data center. The VM traffic bandwidth demand is a Non zero-sum resource that the global traffic sum is relative with each VM placement position. In this paper, we introduce a new improved traffic constant algorithm in the data center, called Degree and Weighted Maximum Traffic Ratio(DWMTR). The proposal DWMTR algorithm defines a new weighted ratio parameter in this paper. The main body of the parameter is constructed with the ratio, current overall intra-cluster traffic divided by current overall inter-cluster traffic, when a new VM places in the data center. The DWMTR algorithm has the ability to constraint the inter-cluster traffic incensement more strictly than the current VM placement algorithms based on traffic bandwidth allocation. For this algorithm based on the theoretical analysis and simulation, it confirms the proposed DWMTR possesses smaller global interactive traffic cost than the control group algorithms in the appointed VM placement in the three-layer data center model.
文摘云计算作为一种新的商业计算模型,提供了弹性计算和存储资源等服务.云计算具有超大规模、高可扩展性、高可靠性、虚拟化、按需服务和价格低廉等特点,许多用户把应用提交到云环境下运行.随着越来越多的数据密集型应用部署到云计算环境上运行,应用会有不同的QoS要求.为了保证在放置应用数据副本时,数据副本都能满足应用的QoS要求,我们需要考虑云计算环境下QoS约束的应用数据副本放置问题,为此提出一种QoS感知的副本放置算法(QoS-Aware Replica Placement with PSO,QRPPSO),该算法首先对QoS约束集的属性进行量化,然后构建目标优化模型,最后在副本放置优化算法引入粒子群优化来决定副本放置位置.仿真实验结果表明所提策略能有效解决QoS约束的应用数据副本放置问题.
