Calling EIGRP a "Hybrid" protocol is Cisco marketing because distance vector
protocols have a bad reputation. Like Al Gore is a "New Democrat" and not
a "Liberal". I like the term "advanced distance vector protocol" to
describe EIGRP because it does do some really interesting calculations such
as finding a "feasable sucessor" route in case of route failure.
The split horizions issue is different with OSPF. Remember, OSPF does not
pass "routes" like a distance vector protocol. Instead OSPF sends Link
State Advertisements to describe the network. Each router then builds a
topological database from those Link State Advertisements. So the protocol
is geared towards distributing the LSAs through the network.
When an LSA arrives at the router, the router checks the database. If the
LSA is new, or newer, the LSA is placed in the database and forwarded out
all other OSPF interfaces. Usually you will not forwared the LSA out the
same interface it was received (sounds like Split Horizons huh).
But sometimes you want the LSA to be forwarded out the same interface it was
received. The behavior may be changed by changing the OSPF network type.
For example, you may be on an ethernet network and the router is a DR. You
would need to make sure the network type is "Broadcast", which is the
default for ethernet interfaces. Or on a hub serial interface connected to
hub and spoke frame relay network you would want to set the network type to
be "multipoint", and not the default of point-to-point. You can find out
which OSPF network type is used on an interface by typing "sho ip ospf
interface".
So, in a nutshell, OSPF does have some concerns that look similar to a split
horizons problem. But the solution is to use the appropriate OSPF network
type and not any command that has the words "split horizons".
Hope this helps!
Paul Borghese
----- Original Message -----
From: <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Thursday, October 12, 2000 6:04 PM
Subject: Re: about OSPF,
>
>
> Just because EIGRP has some of the advantages of a link-state protocol
doesn't
> mean it is, even partially, a link-state protocol. From Doyle, p327...
> "EIGRP is, as the name says, an enhancement of IGRP... EIGRP remains a
distance
> vector protocol... EIGRP is occasionally described as a distance vector
> protocol that acts like a link state protocol". Various other sections
explain
> how it has some advantages of a link-state protocol, but Doyle is very
clear
> that EIGRP is distance vector, not link state, and explains why.
>
> With regards to the original OSPF question, I can't see how split horizon
would
> apply to a link-state protocol. I can't find any reference to an OSPF
> split-horizon command (but if somebody has a URL I'd be interested to see
what
> it says). I'm not positive, but I don't think split-horizon would be
applicable
> to OSPF.
>
> JMcL
>
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
