From: Justin Menga (Justin.Menga@xxxxxxxxxxxxxxxxxx)
Date: Mon Jan 29 2001 - 06:46:33 GMT-3
Yes, not using route maps is simpler, but using route maps is much more
flexible. Route maps are not really complex at all, just following the
config is not quick as it jumps from route-map to ACL and so on. I find the
match and set mentality extremely simple.
Master route maps, they are very important. Of course if you get a
situation that says 'redistribute ALL routes with x metric', then cool, why
use route maps? But if the situation says 'redistribute SOME routes with x
metric, the others with y metric', well sorry, it's route maps.......route
maps are also scalable...
Regards,
Justin Menga CCIE #6640 MCSE+I CCSE
WAN Specialist
Computerland New Zealand
PO Box 3631, Auckland
DDI: (+64) 9 360 4864 Mobile: (+64) 25 349 599
mailto: justin.menga@computerland.co.nz
-----Original Message-----
From: Chuck Larrieu [mailto:chuck@cl.cncdsl.com]
Sent: Sunday, January 28, 2001 4:13 PM
To: CCIE_Lab Groupstudy List
Subject: More thoughts on redistribution options - RIP / OSPF / IGRP
After working Mentor Tech Vlab 2089.
Premise - three routing domains - RIP, OSPF, and IGRP
All subnets are /24
Redistribute such that all routers in any domain can ping all other routers
in any domain
There are redundant paths, so beware of loops and sub optimal routing
Credit it to dumb luck, or maybe this stuff is starting to make sense to me,
but my first time through I managed to construct the redistribution metrics
such that I attained optimal routing ( according to my examination of each
routing table and counting hops against my diagram ) using the basic
redistribute command, and unbalanced metrics. I.e. rip this direction has a
metric of 5, and rip that direction has a metric of 8. OSPF in one direction
has a cost of 10000 and OSPF in the other direction has a cost of 5000, IGRP
this way had a metric of 10000 100 255 1 1500 and IGRP the other way had a
metric of 5000 1000 255 1 1500
OK, time left over. Let's try it with route maps. Whoa, this is not going to
be simple.
What I ended up doing was constructing a set of three access-lists, one each
containing the routes of each domain.
Access-list 1 permit RIP domain routes
Access-list 2 permit OPSF domain routes
Access-list 3 permit IGRP domain routes
Route-map ripside-igrpside permit 10
Match ip addr 1
Set metric 10000 100 255 1 1500
Route-map ripside-igrpside permit 20
Match ip addr 2
Set metric 8000 500 255 1 1500
Route-map ripside-igrpside permit 30
Match ip addr 3
Set metric 5000 1000 255 1 1500
Route-map igrpside-ripside permit 10
Match ip addr 3
Set metric 5
Route-map igrpside-ripside permit 20
Match ip addr 2
Set metric 8
Route-map igrpside-ripside permit 30
Match ip addr 1
Set metric 10
( yes I counted that this would make the max hop count 13 in the rip
tables )
unfortunately, I ran out of time on rent-a-lab before I could construct and
implement similar route-maps and access-lists on the other domain border
routers. So what I saw from placing this set of access-lists and route-maps
on one of the border routers was that all routes were showing up everyplace,
but sub optimal routing galore. I presume this could have been corrected if
I have done similar things on the other border routers.
I bring this up because based on this particular scenario, it appears that
life is far simpler using plain old redistribute commands with a reasonably
careful choosing of metrics rather than trying something fancy like route
maps, which became very complex very quickly.
Just some idle thoughts, maybe to start a discussion or two.
Chuck
http://www.1112.net/lastpage.html
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:27:46 GMT-3