At 4:41 PM -0400 5/13/02, Priscilla Oppenheimer wrote: >At 04:13 PM 5/13/02, Mike Mandulak wrote: > >Lammle refers to EIGRP as being a Hybrid of distance-vector and link state.
This is a confusion caused by Cisco marketing, partially because they associated update-only protocols with a hello subprotocol with link state. That's flatly wrong. > >That's wrong. EIGRP is not link-state in any way. EIGRP calculates a flat >routing table that lists networks, distance, and next hop (distance >vectors). If the list contains multiple entries for a destination (because >there are multiple ways to reach the destination), the entries are sorted >by metric and the one with the lowest metric is selected. This is very >different than how a link-state protocol functions. > >A link-state routing protocol creates a mathematical graph that depicts the >network. A link-state protocol implements a sophisticated process, called >the Dijkstra algorithm, to determine the shortest path to all points in the >graph when the nodes and links in the graph are known. Link-state has a >specific meaning to computer scientists. You'll find a lot of good stuff >about it if you search with Google. A lot of it is not related to routing >protocols. Where link state uses algorithms based on Dijkstra's (which is getting aged and has been modified), first and second generation DV use Bellman-Ford. EIGRP uses Diffusing Update by JJ Garcia-Luna-Alceves, who continues to publish on even more advanced DV algorithms. JJ was not involved in Cisco's EIGRP implementation. > >EIGRP does have some features that make it different from other >distance-vector protocols. Although EIGRP still sends vectors with distance >information, the updates are non-periodic, partial, and bounded. >Non-periodic means that updates are sent only when a metric changes rather >than at regular intervals. Partial means that updates include only routes >that have changed, not every entry in the routing table. Bounded means that >updates are sent only to affected routers. These behaviors mean that EIGRP >uses very little bandwidth. > >EIGRP also determines a feasible successor, which other distance-vector >protocols don't do. Its complex metric is also a feature not found in many >other distance-vector algorithms, (except IGRP of course). > The best descriptions of this are in Alex Zinin's new book on Cisco routing. It's also worth looking at JJ's papers, although they are heavy on the mathematical side. If anybody wants to start getting into the true theory of routing protocols, you'll need at least a general knowledge of graph and automata theory. This is typically an advanced undergraduate course in a CS program, but isn't impossible to learn on your own. -- "What Problem are you trying to solve?" ***send Cisco questions to the list, so all can benefit -- not directly to me*** ******************************************************************************** Howard C. Berkowitz [EMAIL PROTECTED] Chief Technology Officer, GettLab/Gett Communications http://www.gettlabs.com Technical Director, CertificationZone.com http://www.certificationzone.com "retired" Certified Cisco Systems Instructor (CID) #93005 Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=44195&t=43994 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]