Research And Implementation Of Multi-level Optimization Mechanism For Resouce Availability In The Cloud Environment

In recent years,with the development of cloud computing technology and the expansion of market scale,more and more companies and individuals have transferred their business and data to cloud data centers in search of more efficient system operation and maintenance.Data centers are facing huge pressure of the resource load and the risk of service failure.The configuration requirements of massive resources and the high cost of downtime make building a highly available cloud computing platform an important research topic for both users and cloud service providers.In order to achieve the goal of high availability,emerging virtualization technologies such as containers and network virtualization are applied to the management of cloud computing platforms and network resources,providing more possibilities for the optimization of resource availability.At the same time,the high-availability cloud computing platform has expanded the business processing layer and platform management layer of the traditional cloud computing architecture,bringing new resource management and availability optimization challenges.Around the research on resource management of the above-mentioned high-availability cloud computing platform,the existing work has made some progress,but there are still many shortcomings.First of all,in terms of virtualization resource orchestration,most studies focus on single task(container)or virtual machine scheduling,without considering the characteristics of hybrid virtualized computing resources where containers are deployed on virtual machines,and focus on the improvement of resource utility ignoring the availability optimization;secondly,in terms of network resource management,most studies put forward a general network management optimization framework.It is easy to overlook the characteristics of specific network resources such as virtual routers,and many studies have not considered the impact of computing resource load on cloud network performance.In response to the above problems,this thesis studies the hybrid virtualized resource orchestration mechanism based on joint scheduling and the CPU-aware distributed virtual routing optimization mechanism.First of all,in response to the inefficiency of resource management caused by the lack of cross-layer linkage information in hybrid virtualization scenarios,this thesis proposes optimization ideas for cross-layer information linkage,improving resource utility and resource availability,and constructing corresponding resource utility and resource availability.The model abstracts the problem into a multi-objective optimization problem,and realizes the joint optimization of utility and usability with the help of NSGA-II algorithm.Secondly,in view of the problem of network performance degradation caused by high-availability cloud computing network management ignoring the impact of computing load,this thesis takes the placement of virtual routers as the starting point,and analyzes the optimization ideas for virtual router placement through the research on the highly available cloud computing network mode,constructed a binary integer programming(BIP)theoretical model,and designed an approximate optimal virtual router placement strategy based on branch and bound and Newton's method.Based on the optimization mechanisms at the above two levels,this thesis designs corresponding comparative experiments to verify the feasibility and efficiency of the resource availability optimization mechanism.In summary,this thesis designs resource management mechanisms that optimize availability for cloud computing platforms from two levels: business processing and platform management.Considering that most of the theoretical research stays at the stage of numerical simulation,in order to further confirm the feasibility of the optimization mechanism proposed in this thesis and its performance in the actual cloud computing environment,this thesis designs a multi-level cloud computing resource availability optimization system and integrates the two core functional modules of joint scheduling and network resources realize the results of theoretical research.Finally,the deployment and performance index tests in the actual cloud environment proved the improvement of the resource availability of the designed system in this thesis.