云计算环境下作业调度策略研究

发布时间：2018-11-19 08:20

【摘要】：云计算作为一种新型计算模式,向用户提供动态的按需服务,已引起众学者和各公司的关注。作业调度是云计算的关键技术之一,对于满足用户需求和提高云服务提供商服务质量和经济效益具有重要意义。但是,国内外学者对作业调度的研究还不充分,有的研究仅从用户角度考虑,有的仅从云服务提供商角度考虑,优化目标单一;有的从两者角度考虑但考虑因素不全面。同时,接入控制是云计算数据中心对于到来的作业请求是否接收的策略,其作为避免云计算资源过载的一种有效机制,目前的研究还较少。针对上述问题,本文开展了云计算中作业调度策略的相关研究,主要工作如下:1.从用户和云服务提供商的角度出发,以最小化总成本为目标,提出了一种云计算环境下基于改进蚁群算法的作业调度算法。该算法在实现该目标的同时,兼顾了用户的服务质量(QoS),如作业完成时间和所花费用,并考虑了云计算中虚拟机资源的负载均衡。2.对于云计算环境下作业请求数较多、截止时间要求紧且数据中心资源不足的情况,提出了一种基于接入控制的作业调度策略。该策略的主要目标是通过提高作业请求接收数,即作业请求吞吐量,来最大化云服务提供商的利润。同时,该策略使用了效用计算中的惩罚机制、基于优先级的抢占策略、数据中心资源的可扩展性。3.使用CloudSim的内部使用方式修改其源码进行编程,实现了本文所提出的基于改进蚁群算法的作业调度算法和基于接入控制的作业调度策略。在该平台上进行了仿真实验,首先对基于改进蚁群算法的作业调度算法和基于基本蚁群算法的作业调度算法进行了对比,然后对基于接入控制的作业调度算法进行对比验证。前者对比结果表明本文所提出算法总成本较低,且作业完成时间和费用稍低,而且当考虑数据中心中资源负载均衡时,资源负载均衡程度也有一定的改善;后者对比验证了惩罚机制和抢占策略的有效性,即增大了作业请求吞吐量且提高了云服务提供商的利润。
[Abstract]:Cloud computing, as a new computing model, provides dynamic on-demand services to users, which has attracted the attention of scholars and companies. Job scheduling is one of the key technologies of cloud computing. It is of great significance to meet the needs of users and improve the quality of service and economic benefits of cloud service providers. However, the research of job scheduling at home and abroad is not enough, some of the research only from the perspective of users, some from the point of view of cloud service providers, the optimization goal is single; some from the perspective of both but not comprehensive considerations. At the same time, access control is the strategy of cloud computing data center whether to receive the incoming job request or not. As an effective mechanism to avoid cloud computing resource overload, there are few researches at present. In order to solve the above problems, this paper studies the job scheduling strategy in cloud computing. The main work is as follows: 1. From the point of view of users and cloud service providers, a job scheduling algorithm based on improved ant colony algorithm in cloud computing environment is proposed to minimize the total cost. The algorithm not only achieves this goal, but also takes into account the user's QoS (QoS), such as job completion time and cost, and takes into account the load balancing of virtual machine resources in cloud computing. 2. A job scheduling strategy based on access control is proposed for the situation that the number of job requests is large the deadline is tight and the data center resources are insufficient in the cloud computing environment. The main goal of this strategy is to maximize the profit of cloud service providers by increasing the number of job requests received, that is, the throughput of job requests. At the same time, the strategy uses the penalty mechanism of utility calculation, preemption strategy based on priority, and extensibility of data center resources. The internal usage of CloudSim is used to modify its source code for programming. The job scheduling algorithm based on improved ant colony algorithm and the job scheduling strategy based on access control are proposed in this paper. The simulation experiments are carried out on this platform. Firstly, the job scheduling algorithm based on improved ant colony algorithm and the job scheduling algorithm based on basic ant colony algorithm are compared, and then the job scheduling algorithm based on access control is compared and verified. The results of the former show that the total cost of the proposed algorithm is relatively low, and the job completion time and cost are slightly lower, and the resource load balancing degree is improved when considering the resource load balancing in the data center. The latter verifies the effectiveness of the penalty mechanism and the preemption strategy, that is, increasing the throughput of job requests and increasing the profit of cloud service providers.
【学位授予单位】：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】