Improvement of Border Gateway Protocol Against Failure on Autonomous Systems

Abstract

Border Gateway Protocol (BGP) is used for routing among autonomous systems, its main advantage among the other routing protocols is its stability and its ability to maintain and contain large number of updates in its routing table, and so this is very important as its main usage of it is routing in the internet. These autonomous systems create a large number of updates that cannot be maintained/handled by other routing protocol except BGP. BGP has a big problem which is convergence delay due to large number of updates that may reach to minutes in cases of topology failure or problems, also it need to high processing in the CPU memory which may lead to a freezing node in lot of cases. So it's so important to pay more work and attention to reduce these effects that lead to routing instability and nodes freeze. A lot of studied work on solving this problem create a dynamic model change in the BGP routing MRAI (minimum route advertisement interval) according to the network size. This MRAI is a main functional factor to decrease the convergence delay in much networks, but no one study the effect of many flapping node (neighbor) in a topology on the convergence delay and number of created messages which recreated more and more every time the flapping happened. In this paper, the effect of flapping node on different network sizes with various sized failure studied and proposed a solution with good performance in less processing time and convergence delay.