Public internet routing registries (IRR) evolution

Abstract

Internet Routing Registries(IRR) have been around for quite some time now[1] with the sole purpose of providing the place for service providers to store their administrative, routing policy information which can be used in case of BGP malicious/misconfiguration events. Are there any useful service providers policy data stored in IRR? What current limited research has able to answer is that "Quality" of IRR databases is not known". By "Quality" we mean validity of Internet Number Resources e.g. IPv4, IPv6, AS Number registration, routing policy registration, etc by different network service providers in IRR. We have tried to answer this question by looking into the public IRR datasets of approximately last 4 years [2006--2010]. We have found out that current IRR datasets has a lot to offer than its known/practiced i.e. IRR has approximately 50k full peering available. We are investigating how many peering are in harmony with what BGP is announcing and also which is currently published in well known topology datasets like UCLA IRL[25]. As we believe that if accurate peering can be extracted from IRR than they can provide number of new links which are missing in Internet Topology datasets. It can also reduce the usage of active measurements which in itself is burden on the network. We are also designing BGP Security framework based on IRR which will more accurately perform origin AS authentication as well as inferring the complete policy(what is stored in IRR) of AS.