一种面向云多层应用的费用高效的资源管理方法

由于部署方便、维护简单并且不需要搭建自己的私有机房,云数据中心正成为大多数互联网公司 尤其是初创公司和中小规模公司 部署应用程序的首选。在以基础设施为服务的云环境里,互联网公司可以根据应用程序的需要动态租赁云基础设施,从而节省预算开支,并保证应用性能。然而,在现有的业界实践中,云服务提供商提供的负载均衡和资源伸缩服务只能监控虚拟机的使用状态,并不能监控应用程序的运行状态,因此无法准确根据应用程序的服务需求自适应变换资源规模。同时,现有的文献和实践中,也很少有 研究从云基础设施使用者的角度出发,为使用者节省基础设施租赁费用或高效使用已租赁基础设施。据此提出了一种面向基础设施云环境下多层应用的费用高效的资源管理方法,其在降低用户费用的同时,还能充分利用所花费用提高应用程序性能。通过仿真对所提方法业界实际使用的方法 进行比较,结果表明所提方法不仅能够提高应用程序的服务质量和服务性能,也能较大地降低公司在基础设施租赁方面的费用。

Cost-efficient Resource Management for Cloud Multi-tier Applications

Cloud data centers are increasingly becoming the first choice for many Internet enterprises,especially the newly formed and small ones,because it is convenient to deploy the applications,easy to maintain them,and unnecessary to build a private in-house infrastructure.In the cloud environments based on infrastructure as a service,Internet companies can rent the cloud infrastructure dynamically in accordance with the service requirements.In this way,they have the chance to save the costs of renting infrastructure with the performance guarantees.However,the existing methods for balancing workloads and scaling resources in the industry practice are only able to monitor the states of virtual machines,not the custom performance metrics such as the number of live application sessions.So these scaling methods cannot decide the exact resource demands according to the service requirements.In addition,there are few academic studies on reducing the renting costs and employing the billing resources efficiently in the cloud environment.On the basis of these points,this paper proposed a cost-efficient resource management approach for the cloud multi-tier applications,aiming to save infrastructure costs and improve the application performance with the billing resources.Last,we compared the proposed method with the algorithms used in practice by simulating them with the ripe benchmarks.The results indicate that our method can not only improve the quality and performance of the application,but also reduce the cost of renting infrastructure largely.