Research On Performance Monitoring And Load Analysis Of Cloud-scale VMs Based On Application Service

Cloud Computing is a new distributed application system designed to integrate enterprise data center or between multiple data center resources as a service over the Internet to provide users with the infrastructure (Infrastructure as a Service, IaaS), Platform (Platform as a Service, PaaS) or software (Software as a Service, SaaS).A large number of VMs are running in cloud environment to provide Cloud Service. Performance monitoring and Assessment of the VM’s load is an important part of the cloud platform. It has an important significance to improve the quality of Cloud Service. And the VMs,which provide the Cloud Service usually have a wealth of load characteristic. They can be used to identify VM application service program behavior. Therefore, researching the cloud-scale VM performance and evaluating the load is very meaningful.The thesis analysises and discusses cloud computing,virtualization technology, the basic principles and the architecture, IaaS platform generic architecture model deeply. To design and implement the laaS platform performance monitoring and the VM load evaluatiing model based on OpenStack platform.As the number of VM in cloud environment is too large, it is difficult to obtain the VM information, we design a virtual machine load information collection module. It can monitor the VM status and obtain load information effectively. And this thesis proposes optimized kmeans clustering algorithm based on a specified type classification algorithm to calculate the Load Characteristic Vector of the VM. Using the VM information to define a virtual machine which can be divided into:CPU, Memory, or Metwork I/O intensive type. In order to let user or admin to know the performance of the VM, and to calculate the characteristic value through the formula.It is useful for user to take full advantage of PM and VM resource..At last, testing the function and performance of the laaS Monitoring project,and summarizing the work achievement and the future work.