Research On Data Center Structure And Scheduling Mechanism In Cloud Computing

Since the emergence of Cloud Computing, with the continuous development of science and technology and advance by academia and industry, the applications of Cloud Computing are going on developing. Cloud Computing is moving from theory to practice.With the development of Cloud Computing, data center is also improved. Nowaday, data center is not only a site which manages and repairs servers, but also a center of many computers with high performance which could compute and store huge data. Cloud Computing data center contains hundrands and southands, even million of servers or PCs. It has many hetergenious resources. How to design the network structure of Cloud Computing data center is a considerable problem. Data center is a key to promise high scalability and resource usage of Cloud Computing. In addition, replica is introduced into data center, which is an important method to improve availability and performance. It covers single point of failure, poor fault-tolerance, low access performance, and so on. However, replica also causes several problems as follows: consistency of replica, load balance, hardware and communication cost, and so on.This dissertation analyzes and summarizes new characteristics of Cloud Computing data center in deep, and focuses on three main issues: the data center scalability issues with green energy, policy of replica, and scheduling mechanism. The main research contents of the dissertation are as follows:First of all, the concepts of Cloud Computing and data center are introduced. Aiming at the lack of traditional data center, it focuses on the new features of cloud computing data center. It analyzes three important issues in deep: the scalability and green energy issues of the data center, policy of replica, and scheduling mechanism.Second, aiming at scalability and green energy issues of Cloud Computing data center, it makes full consideration of new characteristics of Cloud Computing data center, referrs to well-known data center, and presentes a new data center network structure according to famous Koch Curve, that is the Snow structure. It makes full account of the data center's scalability and low proportion of switches and servers, and can achieve routing within a shorter average path and smaller network overhead.Again, aiming at replica issues in the Cloud Computing and scheduling problems based on replica, it proposes a scheduling model based on replica. Then, by introducing market mechanisms, it presents a price-driven adaptive replica strategy by considering geographic features, network status, and application characteristics. The strategy is adaptive to copy, destruct and tansfer a replica for different applications in order to achieve load balance. Based on the model and data center network architecture, a scheduling strategy is proposed. The experiments show that, it could effectively reduce transfering time for copying replica, and the mean path length is just a half of that of random algorithm.Finally, focusing on scheduling mechanism, to increase providers'revenue, from resource providers'view, this dissertation makes fully account of priority, deadline, profit, and risk, and proposes separately online and batch scheduling policis. Simulation results show that the proposed mechanisms effectively reduce cost of services for providers, increase the total number of completed tasks and profit of providers, and promote the harmonious development of scheduling environment.This dissertation focuses on Cloud Computing data center, policy of replica, and scheduling mechanism, and makes deep research around such three issues. It not only proposes network structure of data center, policy of replica and scheduling algorithms, but also provides a strong foundation for other related researches on Cloud Computing in the future.