From: Chua, Parry (Parry.Chua@xxxxxxxxxx)
Date: Sat Apr 27 2002 - 23:29:42 GMT-3
Your point 1 have problem, 147.8.72.0/22 belong to ethernet of R2, and you crea
te two subnet
as 147.9.72.0/23 and 147.9.74.0/23 in IGRP, ie R6. This is incorrect. When R2
ethernet is down, these network is no longer reachable, but the IGRP still yhin
k that it is reachable,
You can try this, create two secondary ip address at R2 ethernet as 147.8.72.2/
23 and 147.8.74.2/23. This will advertise to OSPF and then redistribute into OS
PF.
For your point 2, you can use tunnel interface, which has less problem then sec
ondary address.
Regards
Parry
-----Original Message-----
From: yuen me [mailto:yuen_me@hotmail.com]
Sent: Sunday, April 28, 2002 12:09 AM
To: ccielab@groupstudy.com
Subject: Secondary address for OSPF IGRP redistribution - focus on this
thing at the mom
I know this topic has been beaten to death. But I want to focus on solution
of secondary address ONLY now to validate its legitimacy as one of the trick
in the bag. Just put aside ip default-network, area range, tunnel,
summary-address at the mom. I want to focus on short mask in OSPF to longer
mask in IGRP. The following diagram illustrates my observation. e0 of R2
belongs to OSPF 10
R2--OSPF 0-147.8.98.0/23--R6----IGRP 147.8.100.0/23------R1
|
OSPF 10 e0 147.8.72.0/22
I have seen two variation of secondary address so far:
1) Break down the 147.8.72/22 into .72/23 and .74/24 and put these two under
R6 iGRP interface as secondary addresses. Nothing changed on R1
2) Create a new /22 secondary addresses under r6 and r1 IGRP interfaces e.g.
147.8.128.0/22
i spot some disadvantages and want someone to validate and comment on them
Disadvantage of 1)
- IP split hortizon is off. Apply distribute-list to overcome it
- R1 does see the .72/23 and .72/24 but will never be able to ping the host
on R2 e0. it is because R6 secondary address is more specific and longest
match rule prevails. Ping packets never get to R2.
So I need to build policy routing on R6 IGRP interface to overcome it. But
then if I need policy routing, why don't I put policy routing on R1 right
away and forget secondary address on R6.
Disadvantage of 2)
- IP split hortizon is off. Apply distribute-list to overcome it
- R1 does receive 147.8.72.0 but it takes on the primary mask /23 (I have
tried 12.1.5T and .9T. Caveat?). then R1 cannot ping the host on R2 e0
belonging to .74.0/23
In conclusion, secondary address is not viable to short VLSM mask to long
fLSM mask scenario. 2) is viable vice versa. when i say viable, that means
not only R1 can see the route but it can ping any host on it.
Am I right ?
Yuenme
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:58:20 GMT-3