From: Xu, James (james.xu@xxxxxxx)
Date: Fri Feb 15 2002 - 19:39:51 GMT-3
I take that for compliment :-).
Thanks for you and everyone's response, and that is the reason why this
Forum is so great! Since the thread started, I have been receiving
overwhelming response. Some negitive, mostly positive. That are all O.K.,
what matters is through all of this discussion, the truth will comes out,
the simple truth about how the redistribution really works in this Cisco
world.
Enough for that. Back to the topic.
Mike: If I understand you corrctly, then
A route from routing protocol can be counted as "eligible for the RIB", if
they would be in the routing table without intervention of another routing
protocol.
I will do some testing in this weekend to see if this added condition will
be enough to explain the redistribution mechanism.
Howard: The reference material is great, that will be helpful for me to get
a good handle for this topic. As a fact, I just scan through the OSPF book
by John Moy, it does offer some good information about route propagation in
a single protocol. However, it stays short to touch the topic of interaction
between two protocols. By the way, do you have links to the free Zebra code,
and GateD code (I believe GateD is a commercial product from NexHop?).
James
-------------------------------
Consider the following:
interface loopback 0
ip address 192.168.1.0 255.255.255.0
!
router rip
network 192.168.1.0
!
router eigrp 1
network 192.168.1.0
!
router ospf 1
network 192.168.1.0 0.0.0.255 area 0
When we look at the RIB (sh ip route,) we see that because routes sourced
from connected interfaces have the lowest AD, only they actually make it to
the RIB.
But each of the routing protocols will contain 192.168.1.0/24 as a valid
route that _could_ be chosen for the RIB even though in this case it wasn't
actually placed in the RIB. And thus, 192.168.1.0/24 is redistributable by
all of these processes.
Mike
----Original Message-----
From: Michael Davis [mailto:miked@netrus.net]
Sent: Friday, February 15, 2002 1:22 PM
To: Howard C. Berkowitz; ccielab@groupstudy.com
Subject: Re: How route redistribution EXACTLY works --- need
confirmation
Great thread James. I like questions that challenge assumptions. What is
it they say about assumptions?!
Howard, I think, as you said before, we won't know definititively how IOS
does this without a thorough review of souce code. Though I'm sure Cisco
won't be sharing those details with us. ;-) I've just tried to describe,
from a pragmatic standpoint, the behavior I've experienced.
Your mileage may vary!
----- Original Message -----
From: "Howard C. Berkowitz" <hcb@gettcomm.com>
To: <ccielab@groupstudy.com>
Sent: Friday, February 15, 2002 12:22 PM
Subject: RE: How route redistribution EXACTLY works --- need confirmation
> >Mike:
> >
> >I like the condition you added, which is "routes eligible for the RIB". I
> >think, you mean another routing source, which has lower adminstrative
> >distance, didn't inject the same route into routing table.
> >
> >For ospf, when you say the routing bit set, do you mean the routes
resulted
> >from SPF, and also in the RIB? In other words, any route which does not
> >exist in the routing table will not be redistributed?
>
> That is definitely the case for BGP; it's part of loop prevention.
> BGP will also not announce a route that has a next hop that cannot be
> resolved in the RIB.
>
> I don't know definitively if that's done in Cisco's IGPs, but, in
> general, it is a basic principle of avoiding loops.
This archive was generated by hypermail 2.1.4 : Thu Jun 20 2002 - 13:46:24 GMT-3