From: Michael Le (mmle@xxxxxxxxxxxxxxxxx)
Date: Wed Feb 14 2001 - 03:40:36 GMT-3
David,
I beg to differ with Devender in this respect because I would use
distribute-lists. He is correct that, in your specific example, it wouldn't
be needed because External EIGRP wouldn't be used to populate the IP table.
But when you're talking about redistribution in general, most of the other
protocols make no distinction between redistributed routes and internally
generated ones.
Take this scenario for example:
10.10.0.0/16 [R1]---[R2]---[R3]
\ |
\ |
[R4]
Let's say R2 is doing redistribution between RIP and IGRP. R2 is learning
10.10.0.0/16 via RIP from R1. R2 redistributes this out to IGRP and R3 and
R4 learn this. R3 tells R4 about that route and vice versa. They in turn
tell R2 about it. Split-horizon doesn't help here because they're both
advertising the route learned from each other, not R2. R2 now has two routes
to 10.10.0.0/16, but IGRP has a lower admin distance. So it will populate
the routing table with this entry instead of the real one from RIP.
I suggest to read Doyle a few more times again, especially the
redistribution chapter. If you don't understand that, you will need to get a
stronger grasp on the routing protocols and how they choose and advertise
routes. It is fundamental and imperative that you know that for the lab.
Michael Le
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
Devender Singh
Sent: Tuesday, February 13, 2001 6:53 PM
To: David Ankers; ccielab@groupstudy.com
Subject: RE: Redistribution
Hai David,
I personally think we should not be using distribute-list. Router will only
redistribute the routes which make to the routing table, by respective
routing protocols. In this scenario, even if we turn the eigrp split horizon
on at some remote router for eigrp, external eigrp distance will make it
worst choice to most of the routing protocols anyway. for OSPF spf will
take of the matters. I personally will not use distribute-lists unless I
have to use them for avoiding loops/dealing with split horizon issues etc or
if am specifically asked to do so.( Bit fuzzy where and where not to). Does
this make sense to you.
Devender Singh
BE(Hons), CCNP
IP Solution Specialist
-----Original Message-----
From: David Ankers [mailto:d.ankers@chello.nl]
Sent: Wednesday, 14 February 2001 1:19
To: ccielab@groupstudy.com
Subject: Redistribution
My dearest lab mailing list,
Badly in need of help with a topic that keeps comming up here, call me
stupid
(I am) but I can't get this at all, I've read Doyle a couple of times and a
load of other stuff and I think every time I understand it but when it comes
to a lab senario I make a complete mess of it. I either miss it off or do it
when it's not needed.
Mutual re-distribution between routing protocols. For me this is a biggy and
there are so many gothas.
What's the best plan of attack for redistribution? Say on the lab I have two
domains one EIGRP and one OSPF, both on the same router. What my strategy
for
the lab would be is to always filter one domain from the other. e.g.
OSPF domain = 10.10.0.0 /16
EIGRP domain = 10.20.0.0 /16
(Written with out aid of a router so the syntax might be off.)
R1
router ospf 100
redistribute eigrp 20 subnets
network 10.10.0.0 0.0.255.255 area 1
distribute-list 1 out eigrp 20
!
access-list 1 deny 10.10.0.0 0.0.255.255
access-list 1 permit any
!
router eigrp 20
no auto-summary
default-metric (blah blah)
redistribute ospf 100
network 10.20.0.0 mask 255.255.0.0
distribute-list 2 out ospf 100
!
access-list 2 deny 10.20.0.0 0.0.255.255
access-list 2 permit any
The key question here is I have a feeling that I don't need to do this as
the
protocols won't feed back routes in this way? EIGRP uses split horizon so
that would stop feed back in one way.... What about OSPF though or say the
other protocol was ISIS? How does the IOS handle it? As you can tell I'm
very
confused about this issue and I always have been. This is a basic issue and
there is no avoiding it, if I filtered in this way on the lab and I didn't
need to, although the config will work (I guess) the procotor will see for
sure that I don't understand this and deduct a lot of points.
The other thing with redistribution is classless to classful, there are some
gothas here as well... Any rules to watch out for?
I guess it doesn't help that I can't get to my home lab now as my wife has
filled my study room with apples. (Refers to an earlier and very stupid
thread.)
D.
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:28:48 GMT-3