云环境下应用敏感的虚拟资源动态调度技术研究与实现

发布时间：2018-03-26 16:17

本文选题：云计算　切入点：资源调度　出处：《北方工业大学》2017年硕士论文

【摘要】：云计算作为一种新的计算模型,利用分布式计算、虚拟化等多种技术将大规模的硬件资源整合成资源池,使用户能够随时随地按需通过网络的形式访问计算资源,为了最大限度的利用云计算平台,提升云资源的利用率,寻找优秀的资源调度策略是云数据中心需要解决的重要问题,在研究资源调度策略时应充分考虑应用的特点及云计算的实际场景,使云平台能够在达到负载均衡的条件下提高资源利用率,从而来制定合理的资源调度方案。本文通过研究云计算的相关技术,深入对现有的资源调度方案进行分析,然后提出了应用敏感的虚拟资源动态调度方案,本文的主要工作如下:首先,我们对应用的类型予以区分,定义了应用敏感度,区别于传统的资源调度,只要应用缺少资源,就盲目的增加各种资源的方式,资源调度应该根据应用的特点,对应用所需要的特定资源进行分配,并且在云平台有多个应用需要进行资源调度时要有一定的调度优先级,从而整体提升资源的利用率。其次,传统的资源调度仅仅通过对资源的监控来调度资源,缺少了资源就增加,多了就减少,或者干脆一次性分配足够的资源,这样虽然能够满足应用在整个运行过程中的需求,但必然会导致浪费,针对这样的问题,我们提出了基于ARIMA预测的资源调度模型,根据资源的需求情况提前进行资源的调度,从而防患于未然,既在一定程度上避免了资源调度的延迟性,也避免了资源过度浪费,提升了资源的利用率。最后,本文实现了一套基于应用敏感的资源调度机制的系统,本文中所设计的模型与算法均已应用到该系统中,目前系统已经通过第三方测试,并完成了交付使用。相比传统的资源调度方式,我们提出的方法能够及时有效的调度资源,并能够提高资源利用率,降低云数据中心的资源消耗与管理成本。
[Abstract]:As a new computing model, cloud computing integrates large-scale hardware resources into resource pools by using distributed computing, virtualization and other technologies, enabling users to access computing resources anytime, anywhere and on demand through the network. In order to maximize the use of cloud computing platform, improve the utilization of cloud resources, and find an excellent resource scheduling strategy is an important problem that needs to be solved in cloud data center. When studying the resource scheduling strategy, we should fully consider the characteristics of application and the actual situation of cloud computing, so that the cloud platform can improve the resource utilization under the condition of load balance. In this paper, through the research of cloud computing technology, the existing resource scheduling scheme is analyzed, and then the dynamic scheduling scheme of virtual resources is proposed, which is sensitive to virtual resource scheduling. The main work of this paper is as follows: first, we distinguish the types of applications, define the application sensitivity, different from the traditional resource scheduling, as long as the application is short of resources, blindly increase all kinds of resources. Resource scheduling should be based on the characteristics of the application to allocate the specific resources needed by the application, and there should be a certain scheduling priority when there are more than one application in the cloud platform, so as to improve the overall utilization of resources. Traditional resource scheduling only through the monitoring of resources to schedule resources, the lack of resources will increase, reduce the number of resources, or simply allocate enough resources, although this can meet the needs of the application in the entire running process, But it will inevitably lead to waste. In view of this problem, we put forward a resource scheduling model based on ARIMA prediction, which can schedule resources ahead of time according to the demand of resources, so as to prevent trouble in the future. It not only avoids the delay of resource scheduling to some extent, but also avoids the excessive waste of resources, and improves the utilization of resources. Finally, this paper implements a system based on application sensitive resource scheduling mechanism. The model and algorithm designed in this paper have been applied to the system. At present, the system has passed the third party test and completed the delivery. Compared with the traditional resource scheduling method, the proposed method can schedule the resources in a timely and effective manner. It can improve resource utilization and reduce resource consumption and management cost of cloud data center.
【学位授予单位】：北方工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.09

【参考文献】