From: Steve McNutt (lpd@xxxxxxxxxxxxxxxx)
Date: Tue Nov 14 2000 - 20:37:45 GMT-3
This email gives away key issues regarding ccbootcamp lab 7. if you haven't
done that lab but are planning to, you should not read this.
This is regarding mutual redistribution at multiple points. earlier today I
was re-doing CCBootcamp lab 7 as it really gave me trouble the first time I
did it awhile back, and I wanted to see how much I have learned.
Anyway, I came up with a solution that seems ok to me, but I though I would
check with the group and see if I've got my head screwed on straight on this
one.
here is the diagram:
[R3]
___ |s0.1,s0.2
e0| ______|______
| s0 { 306 | 206 ) s0 e0
|--[R6]---{-----|-----}----[R2]---|
| |s1 { 602 203 } |s1
| | ------------- |
ISDN | |
| |s0 |s0
|--[R5] [R1]
|e0 |e0
| |
-----------|--------------
|e0
[R4]
The network uses 10.x.x.x/16 with the exception of two loopbacks on r3(not
shown) and the e0 of e2, with is a /24. there is a 0/0 route to r2's e0
which has been redistributed into rip and ospf.
r3 and r6, are on one subnet in the frame cloud, r3 and r2 are on a
different subnet in the frame cloud.
r6,r3 and r4 run ospf. r1,r5,r1 and r6 run rip. routing is disabled on r4.
r6 and r2 are performing mutual redistribution.
my solution for handling redistribution on the ospf side was to bump the
admin distance on OSPF external routes using the distance ospf external
command, since all of the rip sourced routes are going to be e1 or e2 in the
ospf domain. To ensure the r3 was using the best path to the r6-r5 and r2-r1
links I used a route map on the redistribute command to tweak the metrics
for those two subnets.
on the rip side I just bumped the metric up a little on the redistribute
ospf command.
the solution produced an optimal routing environment with minimum effort.
However, Marc russell's solution is to set the default admin distance for
both protocols to be the same, and then use acls to explicitly lower the
admin distance of routes sourced from the native routing domain. when/why
would that be a better approach than my solution? are there pitfalls to my
solution? comments are appreciated.
-s
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 08:25:45 GMT-3