LISP Part 1: Problem Statement, Architecture and Protocol Description
Google Tech Talk
February 10, 2010
Presented by Dino Farinacci.
We will describe the initial problem statement LISP was created for. Since fall of 2006, when the IAB held a routing workshop in Amsterdam, we have found many more use cases for the level of indirection LISP brings. LISP is taking the overloaded semantics of the IP address, where a network device's identity address and location address are separated so one can keep one of the addresses fixed and while changing the other. This first part of a 3-part series will explain the problem statements, provide an architecture deep-dive of the idea, and illustrate how the LISP protocols are used. This session is necessary prerequisite for LISP Part 2 and LISP Part 3.
Dino originally joined Cisco in spring of 1991 and was one of the first two Cisco Fellows. He has built routers for 27 years. Dino currently works in the Data Center Business Unit at cisco where his focus is on building a next-generation platform and operating system for Enterprise and Data Center environments. This platform is the Nexus 7000 running NX-OS which shipped in April of 2008.
His expertise specializes in routing protocols where he has intimate knowledge and implementation experience with IS-IS, EIGRP, OSPF, BGP, IGMP, PIM, and MSDP, as well as IPv6 and MPLS protocols.
He is an advocate for modular operating systems. Dino also has been a member of the IETF for 19 years making many contributions over this period of time. Dino has worked for cisco since early 1991 but was away for 5 years at Procket Networks where he help build the highest speed and most dense router (still to date) in a half rack chassis which ran a fully modular operating system. He has been back at cisco for 5 years where he is currently working on new multicast routing technology such as Multicast Fast-Reroute, AMT, Multicast Virtualization, and layer-2 multicast for Data-Center Ethernet. Dino invented OTV with his routing colleagues in DCBU and wrote the initial implementation on NX-OS.
Dino is not just a multicast bigot but works on many other protocol and OS initiatives. For example, recently he is prototyping an idea called LISP to separate an IP address into an ID and Locator to allow the Internet to scale better. LISP has been accepted as a working group of the IETF where Dino participates intimately authoring 7 Internet Drafts.