| In recent years,the development of computer technology has evolved from the original Bare-metal age for physical devices,to the age of resource-oriented virtualization,to the now-developed and used service-oriented cloud computing age,making virtualization technologies more mature.Virtualization technology has evolved from pure software support to hardware-assisted virtualization,and its application platform has also evolved from traditional virtual machine-level virtualization to lightweight virtual machine level.Today,the different levels of support for underlying hardware between platforms,and the different characteristics of virtual machines and containers(lightweight virtual machines)have caused traditional virtual machines and lightweight virtual machines to occupy a part of the market,which has become the basic use demand of major enterprises and individual users,and has continued to flourish.However,there is a degree of doubt among users and businesses about the level of performance support for virtualization technology.Faced with their own needs,users are difficult to choose a platform that meets their requirements and optimizes the use and allocation of resources.After developing their own service products,enterprises also need to self-test the performance of their products or platforms.So,how to test and evaluate the performance of the product,and continue to optimize based on the evaluation results,in order to achieve higher performance breakthroughs.Therefore,in order to solve the above problems,it is necessary to design a set of more common virtualization performance evaluation schemes,and develop an evaluation tool corresponding to the evaluation schemes based on this.At present,the market lacks testing and evaluation systems that can target different virtualization platforms.In the virtualization platform,there are few comprehensive performance evaluation studies on virtual machines and lightweight virtual machines.Most of the performance of virtual machines and containers is only a qualitative description,or a quantitative test of the performance of a certain aspect.Therefore,this thesis relies on national key R&D technology projects,and takes a self-controllable national development strategy as the background,and designs a relatively common virtualization performance evaluation system,including user load simulation,indicators,scenarios,and tests for the virtualization platform.The design of the process and evaluation model,and represented by the comparison of the virtualization performance of the domestic virtualization platform and the x86 platform,developed an automated and visualized virtualization performance evaluation tool for the system.Enterprises or individual users can choose the test items of the test platform,and simply submit configuration information,then start the entire evaluation process,and observe the real-time changes of certain indicators in the evaluation process on the tool.After the test is completed,you can observe the final test results and charts and review the generated test report.Aiming at the background and significance of this topic,this thesis first investigates the current status of evaluation of existing virtualization technologies and virtualization platforms at home and abroad,and based on the relevant theories of virtualization technologies,combined with existing research work,proposed a new set of Virtualization performance evaluation plan.Secondly,a prototype of a virtualization platform evaluation tool was developed.Finally,experiments were performed using the x86 virtualization platform and a domestic virtualization platform as the tested platforms,and comparative evaluation was performed.Finally,the x86 virtualization platform and the domestic virtualization platform were used as the test platform for experiments,and comparative evaluation was conducted.The experimental results were used to find the advantages and disadvantages of each platform in supporting virtualization performance,so as to adjust the performance of the virtualization platform.The experiment proves that the virtualization evaluation scheme proposed in this thesis can quantitatively perform a comprehensive evaluation of the performance of different virtualization platforms. |
