Improved Algorithms for Large-Scale Topology Discovery

There is a growing interest in how to effectively and efficiently perform the topology discovery task in a network-friendly manner. Because the traditional method has generated so much traffic that they will begin to resemble distributed denial of service (DDoS) attacks. In this paper, we first analyze some current algorithms, then proposed and evaluated an improved algorithm, we called it binary search, which can reduce measurement load and increase interface and link coverage. The median redundancy of binary search is less than 2 while its interface coverage rate can almost meet 100%. The key idea is to divide and conquer: we see the path from monitor to every destination as a linear series of interface, then divide those interfaces into two parts, check whether they have been discovered already, if it is not, then concentrate on that part in a recursive way. During that process, we maintain the detailed information of the path with the sets of (interface, interface, pathlength, IsItProbe) which are the stopping rule in the discovery processing.