Research On High-level Cloud Architecture Based On Cloud Server Cluster Reserve Mechanism

Cloud computing, the current hot spot in the field of information, leads to an transformation of theentire business model and an extension of a new business system. For the end user, it declares the advent oflow-cost but super-computing era, bringing far-reaching impact and change for users, industry andgovernments. As the most profound change of the IT industry, cloud computing provides computing,storage and other resources in the form of services, forming a dynamic, scalable resource usage patternbased on virtualization. This pattern changes the computing, storage and other resources into the socialinfrastructure of providing common services to the user, which is different from the traditional of resourceusage pattern based on desktop. Cloud computing realizes the real transparency of information services.and its flexible, efficient, low-cost mode of operation becomes the important engine of promoting the greendevelopment of the21st century's industry.However, the soaring of the proportion of the global carbon emission of the IT industry draws theattention of the experts, especially the data centers, which is the main carbon emission source of the ITindustry. Presumably, the carbon emission of the global data center will reach205.9billion tons until2020,becoming the fastest growing element in the ICT carbon footprint. As the important place of infrastructureproviding cloud services, not only does the data center host huge energy consumption, but also theinfluence of large amount of carbon emissions arouses the concern of related experts. A typical cloudcomputing data center includes thousands of servers with its power consumption dividing into twocategories: active power and reactive power, and the latter at high proportion. At the present, the powerconsumption of reactive servers of the best server design still accounts for65%of the consumption peak.With the extensive application of cloud computing technology, the cost of cloud computing data center isbecoming increasingly prominent and how to balance the costs and benefits of the cloud providers becomesa critical point of solving this problem.From the point of energy, this paper researches the power consumption of the data center. Theprincipal work accomplished is as follows: Firstly, on the foundation of cloud computing technology infrastructure, a high-level cloudarchitecture based on energy-aware is proposed including cloud users/brokers, service scheduling, resourcevirtualization and the physical layer.Secondly, this paper proposes the dynamic allocation strategy based on reservation mechanism of theserver cluster with the green service scheduling layer as starting research point. The data center of itincludes two server clusters: the main cluster module and the reserved module. there is a trade-offrelationship between the task betrayal and power consumption: the task betrayal rate is inverselyproportional to the number of the running services, and the power consumption is proportional to thenumber of the running services. With this strategy's determining the number of the main module, the bestbalance between task betrayal rate and power consumption is arrived.Thirdly, a model of power consumption problem is built with a detailed analysis of the cost function ofpower consumption.Lastly, the cloud simulation tool CloudSim is used for the dynamic allocation strategy simulationexperiments. The results show that the dynamic allocation strategy based on reservation mechanism of theserver cluster can balance the tolerance degree of user and the power consumption of the cloud providers,and the data center achieves the lowest point of power consumption on the condition of receiving the mostrequests with the benefits reaching the highest point.