Research And Implementation Of Distributed Data Center Resource Scheduling Algorithm

In recent years, the concept of cloud computing has been well known by more and more people in computer fields. There are also a variety of cloud computing products. Among them, the applied researches carried out on the level of “Infrastructure as a Service” by the technique of virtualization of physical resources, are the basis of other cloud computing products, that is what resulting in the generation of a large number of distributed data centers. But it is undeniable that in this development situation of cloud platform, the resource/energy management is still in a phase of extensive management, that is what lead to the waste of a huge number of resources and the operators’ bad earnings, it is also a major factor of restricting data center operator’ benefits and cloud computing industry development.In this context, this paper researches the literature of efficient scheduling algorithm, energy-saving algorithm and profit optimization algorithm in cloud computing resource scheduling filed, for their problems exposed such as the cloud service resource definition is not comprehensive, no comprehensive consideration of quality of service and data center benefits, imperfect distributed data center energy model, this paper has done the researches and the experiments, the author’s major contributions are outlined as follows:1. According to the feature of the distributed data center, this paper built a distributed data center energy consumption model. In this model, author mainly considered the data center cooling system and server system, quantified analyzed energy consumption of cooling system reducing and maintaining the temperature, server starting, shutting down, transferring the status and the running and migration of virtual machines.2. This paper built a distributed data center benefit model with the consideration of SLA. In this model, author researched the relationship between quality of services and benefits of data center, introduced service level agreement to ensure the provision of quality services to clients, also specified the punishment when services provided didn’t meet the demands, thus ensure the benefits of both data center and clients, at the same time, author has considered the influence of the data center energy consumption.3. Based on the models and the Genetic Algorithms, author designed the encoding of distributed data centers’ virtual machine and server mapping. By inheritance, mutation, selection, crossover and other operations, this paper implemented the distributed data centers energy cost optimization and profits optimization algorithms.4. This paper has carried out the comparative experiments between the resource cost optimization and benefits optimization algorithms implemented by Particle Swarm Algorithm and Genetic Algorithm. Author compared the running time and the results between those algorithms, after that, it is proven that the algorithm proposed by this paper is a meaningful attempt in distributed data center resource scheduling fields, it is also proven that the algorithm proposed by this paper can effectively reduce the energy cost, ensure the quality of services and increase the benefits of distributed data center.