Research Of Virtual Machine Scheduling In Cloud Computing

With the development and application of cloud computing in recent years, the cloud computing approach which is represented by laaS (Infrastructure as a Service) is more and more prominent with respect to the manner of traditional IT. Government, finance, Internet, e-commerce and other industrial users have noticed that cloud computing brings many advantages such as low cost, high reliability, on-demand use, dynamic expansion, rapid deployment, rich extended service, etc., and begin to migrate part or all of their business to the cloud. This is a big opportunity and also a great challenge to the laaS cloud computing. The administrator needs to have an efficient resource management and scheduling control systems to comprehensive management of resources and reasonable allocation, and ultimately deal with increasingly diverse user needs and more complex application types trend on the cloud computing platform.Currently, there have been a lot of cloud computing platform solutions from different manufacturers and the open source communities on the laaS level, including the VMWare vSphere, IBM’s SmartCloud and open source OpenStack, Eucalyptus, etc. When users provide IT infrastructure services with these heterogeneous cloud computing platforms, there are many outstanding issuees such as bounding to a single product and difficult to extend and integrate. On the other hand, the current mainstream cloud computing platforms usually use random static or simple rotation scheduling approachs to allocate and schedule resources in accordance with the user’s applications. It is easy to cause the load tilt and resource utilization imbalance. Therefore, it is urgent to provide a compatible heterogeneous cloud computing IT infrastructure resource management and to achieve dynamic scheduling to IT infrastructure resources. In response to these problems, this paper carries out the research from the point of virtual machine integrated management and dynamic scheduling. The main work and contributions are as follows:1. A study analyzing the virtual machine management and monitoring technology under vSphere and OpenStack cloud computing platform, which focuses on the analysis and summary of the virtual machine management and control interface such as platform-dependent, language support, invocation and processing. Besides, a feasibility analysis is carried out about the acquisition, storage and retrieval mechanisms of monitoring data.2. Designed and implemented a set of cloud management system which is compatible to heterogeneous cloud computing platforms, and we set up a virtualized service environment for research based on the system. The system shields the differences of management interface between vSphere and OpenStack cloud computing with a unified management interface, achieves the integrated management and monitoring of heterogeneous cloud computing platform can also provide service interface which is independent of product-specific for cloud platform to host application.3. Presenting a virtual machine scheduling method based on the resource status in real time perception. This method makes scheduling strategy with greedy algorithm based on the real-time monitoring of resource usage of virtual machines, and finally through the virtual machine live migration technology to achieve dynamic load balancing and energy optimization of the cloud computing platform. According to the traffic big data processing environment for test cases, the method is proved effectiveness under the strong load regularity of the system.