The Research And Implementation Of Cloud Resource Scheduling Based On Application Container

With the rapid development of cloud computing, virtualization technology as one of the important technology also advancing with the times. Full-virtualization,para-virtualization technology cannot satisfy all scenarios in the cloud computing nowadays. Because virtualization environments build by these techniques tend to have a relatively independent system resources, this strategy caused that applications in different virtual machines is very difficult to achieve interoperability, so container-based virtualization technology appeared. Container solved the problem of the procedure independence and resource sharing perfectly. And resource scheduling is a key problem in cloud computing, how to achieve the efficient allocation of resources, how to distribute the appropriate physical host dynamic for your virtual machine is also a hotspot and difficulty in the field of cloud computing research. Despite the large volume of research published on energy efficient VM live migration in cloud data centers,there are very few open-source software system implementing dynamic VM consolidation. So the cloud resource scheduling based on application of container has important theoretical and practical significance.This thesis includes the following main works. Firstly, it introduced the mainstream virtualization technology, including full-virtualization, para-virtualization,and a lightweight virtualization technology based on container, which is appeared recently(operating system virtualization). Secondly, it studied the commonly used system migration technology, including physical host to virtual machine migration(p2v),virtual machines to physical host migration(v2p), virtual machine to virtual machine migration(v2v). This thesis is focus on dynamic and static migration of virtual machine to virtual machine migration. Thirdly, it analyzed a C function library libvirt, which support the mainstream virtualization tools under Linux. Fourth, it designed the general framework of virtual machine scheduling system.Finally the thesis implemented a cloud resource scheduling framework system based on container. The system has three core models,deciding when to trigger virtual machine migration,selecting virtual machine to migrate,selecting host to placement virtual machine. The system has good scalability,it supports to extend the schedulingalgorithm corresponding to implementation the interface and configuration. In order to verify the validity of the system, the thesis finally implemented a prototype system aiming at energy consumption. And the thesis gives out the testing process and result of the system. The test results show that the system can complete virtual machine migration based on the algorithm which user has configure.