Research Of Key Technology For Building NaaS In Cloud Platform Based On Complex Environment

Cloud computing is an emerging computing resources, storage resources and network resources delivery mode and business models. Using virtualization technology, Cloud computing make the hardware infrastructure virtualization, and then serve the customers in a dynamic and flexible way. But with the rapid development of cloud computing, cloud computing center is also increasing, the network is becoming more and more complex. However, the traditional network architecture is difficult to meet the flexible and diversified resource requirements cloud computing and related business asking for. Abstracting network equipment infrastructure into the service(namely NaaS) which can be delivered to the customers, and making the resource convenient for the user administrator to control and to use, provide the cloud computing platform with an effective and feasible scheme of network architecture. How to manage large and complex network infrastructure equipment effectively in a cloud computing data center? Guaranteeing the quality of service, how to manage physical devices and virtual network from a global perspective? Nowadays, all the problems above are needed to be solved in the cloud computing platform.This paper builds NaaS service on the base of OpenStack cloud platform, and research its internal virtual machine network traffic deeply, then put forward the corresponding solutions. In this paper, the main work is as follows:1. This paper expounds the definition and architecture of cloud computing and software-defined networking(SDN), as well as the solutions of NaaS in cloud computing.2. This paper analyzes the infrastructure of OpenStack and the working principle of Neutron, and establishes a complete, private cloud platform, which includes Neutron service. After analyzing the single point of failure in the Neutron network structure, and contraposing the network traffic of the virtual machine and bandwidth problems, this paper adopts the distributed virtual routing(DVR) technology to improve the network quality of service, so that the problem of network single point can be solved.3. According to the request of internet application obeying Poisson process theory, this paper analyzes the QoS in DVR environment, and presents a simple network model, which can estimate the number of VM in compute node according to the service time and ability of VM. As a basis for the virtual machine cluster expansion or contraction, the model realizes the automation of compute cluster for meeting SLA of users.