"Hybrid" routing protocol is a marketing term, not a technical one,
which is an attempt to differentiate EIGRP from older distance
vector, and standards-based link state, protocols.
EIGRP uses the Diffusing Update algorithm invented by JJ
Garcia-Luna-Aceves while he was at Stanford Research Institute. He
was not involved in Cisco's EIGRP implementation, and indeed has
published several new algorithms he believes faster than DUAL, such
as Link Vector Algorithm with Sequence Numbers (LVA-SEN). JJ, when I
last looked, is on the faculty of USC Santa Cruz.
Simplifying a great deal, distance vector routers exchange their
routing tables (subject to split horizon) with their neighbors, add
their incremental link costs to the routes received, and pick the
best routes. Link state protocols exchange information about routers
and directly connected links, including accurate copies of this
information from non-neighboring routers, and independently create
routing tables.
Distance vector protocols are sometimes called Bellman-Ford or
Distributed Bellman-Ford, while link state are sometimes called
Dijkstra. These names refer only to parts of the algorithm
I tend to think of generations of distance vector protocols:
1st (IP RIP, RTMP, XNS RIP): hop count metric, periodic plus
optional triggered update, loop prevention through split horizon and
basic holddown, loop detection through count to infinity, unreliable
transfer of routing updates
2nd (IGRP, IPX RIP): bandwidth/delay metric, periodic plus optional
triggered update, loop prevention through split horizon and basic
holddown, loop detection through sensing monotonically increasing
metric or count to infinity, unreliable transfer of routing updates
3rd (EIGRP): bandwidth/delay metric, updates on change only,
loop-free route computation algorithm, reliable transfer of routing
updates
BGP's path vector algorithm is a variant on distance vector.
> I just a general question about routing protocols, if anyone could help
>me out here I'd be grateful.
> When comparing EIGRP to Distance Vector routing protocols, like RIP,
>the only similarity that I noticed was that the network statements are both
>classful.
It's an "urban legend" that classful vs. classless has ANYTHING to do
with something being link state or distance vector. It's a
historical accident that the first dynamic routing protocols,
developed when there was no such thing as classless addressing, were
distance vector. Indeed, RIPv2 is fully classless although a quite
old design.
>Is this the only characteristic that prevent EIGRP from being
>considered a total link-state routing protocol? Or is there something else I
>failed to notice?
>
>
>Thanks in Advance,
>Freddy Krugar III
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
