云计算环境下面向关联关系的资源调度器

发布时间：2018-05-28 17:02

本文选题：云计算 + 资源调度　；参考：《西安电子科技大学》2014年硕士论文

【摘要】：在互联网时代,云计算是一种能将大量的计算资源整合,使得用户能够按照需要运行大量的计算任务的新兴计算模式。云计算平台的调度器作为资源整合的核心模块,在优化资源,提高计算任务性能以及降低平台风险上起着重要的作用。随着云计算平台上的任务运行的任务更加复杂,更加多样化,云计算平台的调度器的作用变得越来越重要。随着计算任务的负载增大,云计算平台对虚拟资源的管理成为云计算平台面临的新的挑战。目前的云计算平台对于存在关联关系任务的调度存在着资源利用率低,分配不公平导致负载不均衡,进而影响云计算平台性能的问题。针对如何提高云计算平台资源利用率、资源分配公平性,以及负载均衡问题,提出了面向关联关系的主导资源公平性算法(CBDRF)。主要的研究工作有以下几个方面:(1)针对云平台计算管理资源问题,结合云计算平台的结构,重点分析了资源分配的特点、平台调度模型以及平台调度器。(2)针对云计算平台中关联任务资源分配时公平性不足、资源利用率低等问题。本文采用图论的相关模型,结合最小最大原则以及DRF算法,对云计算平台的资源调度问题进行建模。在此模型基础之上,设计实现了一种基于DRF和图论算法的的CBDRF调度算法。能够在任务关联公平性和执行效率之中平衡,有效的提高了资源的利用情况。(3)在实现CBDRF算法的基础之上,结合云计算调度的研究现状和需求。完成了云计算调度器的设计实现并在Openstack平台上进行了实际的测试验证。
[Abstract]:In the Internet era, cloud computing is a new computing model that can integrate a large number of computing resources and enable users to run a large number of computing tasks according to their needs. As the core module of resource integration, the scheduler of cloud computing platform plays an important role in optimizing resources, improving computing task performance and reducing platform risk. As the tasks running on cloud computing platform become more complex and diversified, the role of scheduler of cloud computing platform becomes more and more important. With the increasing load of computing tasks, the management of virtual resources in cloud computing platform becomes a new challenge to cloud computing platform. The current cloud computing platform has low utilization of resources for scheduling related tasks, which leads to load imbalance, which affects the performance of cloud computing platform. Aiming at how to improve resource utilization, resource allocation fairness, and load balancing in cloud computing platform, an association-oriented dominant resource fairness algorithm (CBDRFF) is proposed. The main research work has the following several aspects: 1) aiming at the cloud platform computing management resource problem, combined with the cloud computing platform structure, the paper emphatically analyzes the characteristics of resource allocation. The platform scheduling model and platform scheduler. In this paper, the resource scheduling problem of cloud computing platform is modeled by using graph theory model, minimum maximum principle and DRF algorithm. Based on this model, a CBDRF scheduling algorithm based on DRF and graph theory algorithm is designed and implemented. It can balance the fairness and execution efficiency of task association, effectively improve the utilization of resources. Based on the implementation of CBDRF algorithm, combined with the research status and requirements of cloud computing scheduling. The design and implementation of cloud computing scheduler are completed and tested on Openstack platform.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【相似文献】