andExperience.PDF

Implementing the Locator/ID Separation Protocol: Design and Experience Luigi Iannone and Damien Saucez† and Olivier Bonaventure† Deutsche Telekom Laboratories AG, Technische Universität Berlin, Berlin, Germany luigi@net.t-labs.tu-berlin.de † ICTEAM – Université catholique de Louvain (UCL), Louvain-la-Neuve, Belgium {first.last}@uclouvain.be Abstract During the last few years, the network research community and the industry have been working on the design of an alternate Internet Routing Archi- tecture aiming at solving the issues arising in the current architecture. It is widely accepted that applying a Locator/ID Separation paradigm would result in a more scalable and flexible architecture. As the name suggests, in Locator/ID Separation the identification (ID) and the localization (locator) of end-points is separated, while the link between the ID and the Locator(s) is ensured by what is called the mapping system. In this paper, we present OpenLISP, an open source implementation of LISP (Locator/ID Separation Protocol). LISP is a Locator/Identifier sepa- ration solution based on the map-and-encap approach, which has the merit of being incrementally deployable, hence, falling in the category of dirty-slate approaches. OpenLISP is not a merely implementation of the LISP specifica- tions; it also defines the mapping sockets, a socket-based abstraction making the data plane and the control plane implementations independent. The evaluation provided in this paper shows the limited impact on the protocol stack performance compared to traditional non-encapsulated trac. Key words: Routing, Architecture, FreeBSD, LISP, Future Internet. 1. Introdu