The Virtual Machine Resource Management Based On Resource Aware In IaaS Cloud

The Infrastructure as a Service technology (IaaS) is one of the main services types of Cloud Computing. For IaaS, the customers are able to lease the infrastructures in virtual data center without constructing the expensive and inefficient self-used data processing infrastructures. However, it means that the economic benefit of IaaS supplier is greatly dependent on the management efficiency for virtual resource.The Virtual Machine Consolidation (VMC) technology is one type of efficient optimization technology for virtual resources.The VMC is able to distinguish the resource utilization sample points series of the given time frame by the Time Series Analysis methods. It can also match and consolidate the resource complemented virtual machines in the selected servers based on the resource utilization sample points series. Thus, the technology can improve the resource multiplexing level, reduce the management complex and cost for IaaS supplier. Because of the less working physical servers, it is in according with the green and low-carbon ideals.However, there are many challenges for VMC in the data center. The most fundamental problems is:how to balance the high quality service level guarantee and the maximum resource utilization demand. The problem may influence the operational efficiency of virtual data center to a great extent.For the problem, it is based on the prerequisite that the resource utilization series of the virtual machines can be forecasted in high precision and confidence level, the dissertation plans five core topics:the one-dimension machines segmentation and iteration match process; the fusion consolidation model based on capacity constraint; the consolidation potential analysis of virtual machines based on the measurable physical quantities; the multi-dimension virtual machines consolidation; the virtual machine placement limited by traffic bandwidth. Furthermore, the five core researching topics can be concentrated into three research subject:The first subject makes the related researches about the one-dimension virtual machine consolidation. It includes the first three core topics:the one-dimension machines segmentation and iteration match process; the fusion consolidation model based on capacity constraint; the consolidation potential analysis of virtual machines based on the measurable physical quantities. The relative achievements are introduced in the Chapter 2.The dissertation researches the fundamental performances of one current pure dynamic consolidation algorithm firstly, named Iterative Correlation Match Algorithm (ICMA). Then, it proposes one improved scheme against the performance defects of the pure dynamic consolidation algorithms. In the scheme, it constructs the process of segmentation and iteration consolidation process based on the peak resource utilization classifying. Next, in the second core topic, it proposes one type of fusion consolidation model which combines the advantages of dynamic consolidation model and static consolidation model. And it designs the forecast systems for the consolidation potential of the single virtual machine in the third core topic.The mentioned three core topics run the research for one-dimension virtual machine in different research fields. Finally, it achieves one type of efficient consolidation algorithm for one-dimension virtual machine, named Segmentation Iteration Correlation Combination.The second subject is introduced in the Chapter 3 according to the fourth core topic. The research subject is relative with multi-dimension virtual machine consolidation. It also designs a efficient consolidation algorithm for two-dimension virtual machine based on the research achievements.The dissertation infers that the inter-dimension interference is the key obstacle for efficient Virtual Machine consolidation synchronously. It defines the conceptions about Inter-dimension Collaboration (IDC) to analyze the performance limit of efficient multi-dimension consolidation model. In the Chapter, it designs the efficient multi-dimension consolidation mechanism based on the IDC analysis and makes a instance for two-dimension consolidation environment. The mechanism calls the Utilization Maximum Consolidation Algorithm (UMCA) for the IDC virtual machines. And it applies the Utilization Acceptable Consolidation Algorithm (UACA) for the Non-IDC virtual machines.The third research subject is given in the Chapter 4 of this dissertation, which is the fifth of the five researching topics. The dissertation studies the problem of traffic bandwidth constraint in the three-layer switch model of the data center. And based on the study, it proposes a heuristic flow allocation model named Degree and Weighted Maximum Traffic Ratio (DWMTR) model. The DWMTR algorithm is pertained to the virtual machine consolidation algorithms mentioned in the first three chapters. The DWMTR. algorithm can minimize the influences to the performance of the virtual machine consolidation algorithms by arranging the virtual machine at the proper placements in the data center.