From: asadovnikov (asadovnikov@comcast.net)
Date: Wed Oct 08 2003 - 01:08:12 GMT-3
I assume it is real life question, as it sound unusual for CCIE
preparation... So you should have some flexibility on what you do.
Although some details are missing from you analyses on a big picture it is
very accurate. So the real question is how to get some IP traffic to take
on link, and the rest of it to take another link, while allowing them to
take the same link during a failure.
If you are running EIGRP it will be real simple. EIGRP allows per-route
metric manipulation, so you can dedicate 2 Loopback interfaces for DLSW (and
nothing else) and make the routing to just this subnets better metric one
link, and the rest better metric on another link.
OSPF however does not have such flexibility, so there is no simple solution
for what you want to do in OSPF based network.
Couple things which may help you though:
Policy routing
--------------
You actually can use policy routing in this scenario. You still want to
dedicate 2 loopbacks, one per router for DLSW. Include this loopbacks into
OSPF. Then you want to unequal cost your parallel links (so normal IP would
take the link with the lower cost), but run OSPF over both links. Then you
make sure CDP is running on both links, and turn the timers way down. Then
you set policy routing (or local policy routing in case DLSW is on the same
routers that terminate the links) to prefer the higher cost link. All of
this you knew already... Now, there is a knob on set command in route-map
"verify-availability" which uses CDP to confirm that next hop is reachable.
http://www.cisco.com/univercd/cc/td/doc/product/software/ios122/122cgcr/fipr
rp_r/ind_r/1rfindp2.htm#1026761
Another routing protocol
------------------------
You still want to dedicate 2 loopbacks, one per router for DLSW. Include
this loopbacks into OSPF (primary routing protocol). Then you want to
unequal cost your parallel links (so normal IP would take the link with the
lower cost), but run OSPF over both links. Then run another routing
protocol (say EIGRP) over link which has higher OSPF cost and include the
DLSW loopbacks into EIGRP, and make EIGRP admin distance better. Then while
EIGRP link is up DLSW goes over EIGRP and the rest of IP over OSPF, if one
fails - all traffic goes single link.
Let me know how these solutions fit into you environment and if they do not
I can come up with something else (even less straight forward :) ).
Best regards,
Alexei
P.S. Depending on your environment you may be able to use DLSW redundancy
features to help the case but from your initial description I can not see
immediate benefit.
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of Tom
Young
Sent: Tuesday, October 07, 2003 11:18 PM
To: ccielab@groupstudy.com
Subject: DLSw's routing (difficult)
Hi, group
I meet a difficult question, I am in HQ, have two
lines connect with remote office.
One line service for IP packet, and another one
service for DLSw, I know I could use route-map to deal it,
but I need the DLSw's backup!
I means if the dlsw line was cut , it could use ip's
line to reconnecting.
The difficult point is: DLSw should use dlsw line to
transfer the SNA packet. And IP line back up it !
You know DLSw is base on IP, it use IP route to set relationship, and use
that line to transfer SNA, So, infact, althought the dlsw peer connected on
dlsw line's points, the SNA packet was still transfered by IP's line. I
confused it.
Thanks alot
This archive was generated by hypermail 2.1.4 : Mon Nov 24 2003 - 07:52:58 GMT-3