Target Detection Technology Research In The New Network Environment

With the rapid development of the Internet, the emergence of new network services and applications has gradually increased the size of the Internet. Therefore, the requirement of the internet in all aspects has enhanced, especially the difficulty of network management. What’s more, the network topology detection is the foundation of network management, it decides whether the network manager has a deep understanding of the network, and whether the network manager can manage the network well.Cloud computing is a new type of internet application raised in recent years. By utilizing the network, cloud computing connects the storage device with the server, in order to create a resources pool to realize the resources sharing. However in the cloud computing network environment, external network cannot achieve the access to the internal cloud environment, besides the virtual machines which is in one calculation node just use the same network protocol. Due to those characteristics of the cloud computing network environment, the method which is based on protocol characteristics or detecting packets time difference cannot obtain the distribution of the internal virtual machine in the cloud computing network environment. Therefore the traditional topology detection method cannot be used in the new cloud computing network environment.This paper presents a research of the topology detection technology which can be applied in the new cloud computing network environment. The mainly completed work includes:1. The analysis of the topology detection method based on the protocol and the tomographic imaging detection method in traditional network environment.2. The proposal of hierarchical topology detection method on the basis of analyzing the existing detection methods. The hierarchical topology detection method divides the new network into two layers, and different topology detection techniques are applied to different layers. Firstly, the tomographic imaging detection method is applied to the No.1 layer which includes the external routers/switches, cloud network calculation nodes and external routes. Secondly, the method which uses the ping detection packets to detect the topology is applied to the No.2 layer which includes the virtual machines that are working on the calculating node inside of the cloud computing network.3. Design of a prototype system based on the above statement. In the No.1 layer, the sandwich packets which are based on TCP and UDP are utilized to realize topology detection. In the No.2 layer, firstly, the ping packets are sent between the virtual machines, secondly the MPI is applied to collect the delay data, thirdly the data clustering is realized by using the union-find. At last, the result proves that the proposed topology detection method is effective.