Fault Diagnosis Method In The Network Virtualization Environment

Network virtualization as one of the basic features of future network has been endorsed by major organizations in the world. In the network virtualization environment, the traditional Internet Service Providers (ISPs) were divided into Infrastructure Providers (InPs) and Service Providers (SPs). The InPs are responsible for building substrate networks and providing physical resources. The SPs lease substrate resources, build virtual networks and deploy protocols and applications. SPs are divided into virtual network providers and application providers in some other papers. The network virtualization shields the details in the physical networks. On the one hand, it offers the possibility to support the various protocols and applications in the networks. On the other hand, it reduces the cost of operations and buying infrastructures by sharing the substrate infrastructures. For the foreseeable future, the virtual networks will support more network protocols and applications. But the network robustness and survivability will become a problem that should not be ignored in the progress of the future network development.The network fault management aims to find network faults timely and accurate, quickly isolate and recover faults so that reduce the impact of network faults in the services. Fault diagnosis as part of fault management is very important.This paper studied the four problems about the probe station selection method and fault location method in the network virtual ization environment. The problems about probe station selection method are as follows:(1) the impact of dynamic mapping relationship on probe station selection;(2) both the network nodes and links may break down, so we need to probe the states of network nodes and links. But the existing probe station selection methods only consider the existence of network nodes. The problems about fault location method are as follows:(1) the information about substrate network is transparent to service providers, which will lead to the substrate faults cannot be found;(2) there are multiple faults in the network virtualization environment.Base on the existing research results of probe station selection method and fault location method, this paper studied the problems about probe station selection method and fault location method in detail. The contributions in this paper are as follows:(1)In order to reduce the impact of dynamic mapping relationship on probe station selection, this paper provides a method in the dynamic network environment. This method also provides convenience for probe selection and reduces the number of probe stations.(2)This paper not only considers to probe the statuses of node faults, but also to probe the statuses of link faults. In order to consider probing the status of link, each link was seen as a node in this paper.(3)This paper provides a multi-layer fault diagnosis method which is based on the features of multiple networks in the network virtualization environment. First, this method infers virtual faults in the virtual networks. Then it infers substrate faults based on the information about virtual faults and the dependency between virtual networks and substrate networks, In this way, it will reduce the influence about dynamic network environment and reduce the cost of fault recovery based on the distinction between correlative virtual faults and independent virtual faults.