Towards a more efficient Internet? Introduction to the Locator/ID Separation Protocol ( LISP)

The growth of the Internet is measured in particular by the number of IP routes that enable access to any machine connected to the network. IPv4 routes represent several hundred thousand entries (in 2016) in the tables maintained by the routers at the heart of the Internet, which have a permanent overview of the network's topology and all the routes that traverse it. Similarly, IPv6 routes account for several tens of thousands of entries (in 2016) in the tables maintained by routers at the heart of the Internet, and we can be sure that this number will continue to grow over the coming decades, as the IPv6 protocol is systematically adopted as the preferred means of communication for machines connected to the Internet, and in particular those making up the Internet of Things, estimated to number several billion by 2020.

Clearly, the growth of routing tables has a significant impact on traffic routing efficiency on the Internet: the founding "longest match" principle on which routers base their IP packet routing decisions requires them to systematically consult all the entries in their routing tables in order to identify the most specific route that will enable them to determine the output interface through which packets destined for a given network will be transmitted.

The more entries in the table, the longer it will take to decide which packets to route.

Despite attempts to obliterate the notion of IPv4 address classes in order to optimize prefix aggregation efficiency when prefixes are to be announced by Internet routers ("Classless Inter-Domain Routing" technique, as specified in RFC 1584), or to facilitate prefix aggregation capability through a fundamentally hierarchical addressing formalism [as introduced by the original typology of IPv6 addresses into TLA (Top Level Aggregator), NLA (Network Level Aggregator) and SLA (Site Level Aggregator)], it has to be said that the average lengths of IPv4 and IPv6 prefixes exchanged between routers on the Internet are not of a nature to control the growth of routing tables maintained by routers belonging to the "Default Free Zone", also known as DFZ, this region of the Internet made up of routers that maintain all available routes and are therefore free of default routes.

LISP (Locator/ID Separation Protocol) is a recent attempt, specified by the IETF (Internet Engineering Task Force), to meet this need to control the growth of routing tables maintained by DFZ routers. This protocol is part of a family of initiatives aimed at significantly improving the efficiency of Internet traffic routing, based on the founding principle of separating information characteristic of the terminal's identity from that characteristic of the network location to which that terminal is connected....