Research On The Key Issues Of Proactive IP Network Survivability

Internet has seen tremendous growth in the past decade with the introduction of new science and technology and the emergence of new services, so that service providers find out that more and more traffics in the Internet have the potential to bring more profits. But the failures happening everywhere and every time even in the robust network may make the service unavailable and performance digressive. Network survivability is popular among service providers to guarantee Qos. The main research content of survivability is a network how to complete the service on time dynamically in face of challenges such as failure and attack. As the pure IP network is exactly the component in real network, the IP network survivability becomes hot because of its lower cost,fine-grained,and longer time to protect. This dissertation don't take Network Security into account.There are two directions: Reactive and Proactive based on the establish time of backup-path. But Reactive Policy wastes so much time on failure recovery time which can't satisfy real-time service Qos requirement such as VOIP. Therefore this dissertation focuses on the Proactive IP network survivability for routing flap suppression mechanism handling frequent failure,network single transient failure recovery mechanism and multi-link-failure solution based on minimum cover set. The main contributions are as follow:(1)The causes of routing flap in real backbone network are analyzed. And the relationship of link flap and routing flap is given based on the determinants at every phases of routing convergence. The suppression of routing flap transfers into the suppression of link flap. A localized approach is proposed to suppress the routing flap through dynamically tuning the Hello timer. Simulation results show that this mechanism can achieve better performance in eliminating routing flap and consume much less computation resources at the same time. And a small network is built to verify it can interoperate with the existing routing protocol.(2)After analysis of the existing failure recovery algorithms handling single failure, a single-link failure recovery mechanism based on backup-link is proposed. It considers about the differences and the relations between SPF of the node failure or not, and picks up backup-link to build loop-free backup-path. This mechanism not only has 100% network coverage and a lower algorithm complexity, but also improves the two-way link limitation of Defection Routing and the dependence of Failure Insensitive on hardware. Routing. Experiments prove the algorithm has better performance.(3) In order to handle the single node failure problem, backup links are used to connect the depart sub-trees of failure node to the other sub-tree of the root node SPF after the analysis of their relationship based on BLSL algorithm. Loosely constrained routing is proposed to transfer data through routers in the backbone network. It is the way to keep loop-free routing in backup-path, and decreases the number of routing table entry at the same time. Simulation results prove that the backup path of this algorithm is almost the same as the shortest path and better than the others.(4)Multi-topology technology which is used to solve the problem of multi-failures is the method to solve minimum cover set problem. Sub-topology routing algorithm is proposed based on the knowledge of Graph Theory about minimum spanning tree and its extended tree. Every sub-topology protects some links which don't belong to it, all sub-topology keep all links safe. The number of sub-topology in a network is depended on the number of the continuous isolated nodes in extended tree which can not prove by the other multi-topology technologies. And experiments certificate that the average backup path length in sub-topology is smaller and it has a lower the packet loss rate.