>"Winchester, Derek S." <[EMAIL PROTECTED]> asked,
>Looking for an educated answer other than the book definitions that explain
>Convergence in Eigrp: What feature or why is Convergence in Eigrp faster
>than other Link-state and distance vector protocols. I would like it in the
>words that actually makes sense. I know that it is faster, and I know that
>Dual makes convergence simpler and fast, but in words of the common man,
>WHY?
It's misleading to say EIGRP is faster, unless you compare it to
other modern protocols with the same hello timer value. By default,
EIGRP has a 5 second hello timer while OSPF and ISIS have 10 seconds.
Obviously, a protocol with a shorter hello timer will detect and
respond faster to changes.
Once the hello timers are equated, will EIGRP, OSPF or ISIS be the
fairest in the land? The mirror on the wall says, "it depends."
Tony Li, who used to be one of the principal routing developers at
Cisco, and is coauthor of BGP, discussed this a while back in one of
the IETF mailing lists, and described EIGRP as faster to converge
when the alternate path was in an adjacent router or one that is one
hop away from the adjacent router, while OSPF is faster when finding
the new path will involve routers more than one or two hops away.
The reality is that in most circumstances, there is very difference
among the real-world convergence times of the three modern IGPs. If
you were doing something such as non-local-acknowledged SNA, which
has critical convergence time requirements, you'd have to tune the
timers of any of them.
EIGRP probably needs less resources, but needs more tuning for low
speed lines (below 56 Kbps, roughly)
RIP and IGRP, however, are significantly slower than the modern
protocols, principally because they (1) have to wait for periodic
updates and (2) usually have holddown timers that make them wait for
several updates.
There is a whole other issue, however, on what do we mean by
convergence? I'm in the process of writing an Internet Draft on some
definitions, especially for BGP.
Is convergence:
The time it takes a router to initialize its routing table on power on?
The time it takes to be able to route to a new (i.e., previously
unknown) destination, after it receives the update? After the change
causing the update occurs?
The time it takes after it recognizes the best route to a
destination is down, but it has an alternate route in local memory?
The time it takes after it recognizes the best route to a
destination is down, but it does not have an alternate route in local
memory?
___________________________________
UPDATED Posting Guidelines: http://www.groupstudy.com/list/guide.html
FAQ, list archives, and subscription info: http://www.groupstudy.com
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
