From: Raymond Adi Irawan (Raymond.Irawan@xxxxxxxxx)
Date: Wed Nov 15 2000 - 02:02:28 GMT-3
I've done this lab before. I'll try to answer your questions. Please anybody
correct me if I made any mistake.
-----Original Message-----
From: Jennifer Joy [mailto:jjoy@tri.sbc.com]
Sent: 15 Nopember 2000 9:07
To: ccielab@groupstudy.com
Subject: bootcamp lab 8, ospf<->bgp questions
Now the WAN link is the only network being advertised via the
EBGP peer. So why the route-map? I am not sure. Is that to
make sure the route is there so we can run without "no sync"?
>>> First, we need redistribute connected into ospf so that R1 has route to
the next hop BGP route from R8. Without doing so,
R1 won't advertise to R7 (BGP rule 1).
We need route-map here, especially in R6. Remember, bri 0 is also the
connected network. Even you've put bri0's network into ospf, ospf will also
treat it as Type-5 external route. Looks like doesn't make sense but you can
see it using show ip ospf database. If we don't put the filter, your isdn
link will keeps flapping. Here's the logic: ospf demand circuit will up if
there's any topology changes. When bri0 goes down (idle-timeout expires),
ospf will learn the topology changes (removing bri0 Type 5 external route)
so it will start to flood LSA including to the bri0 itself. This will make
bri0 goes up and down again after idle-timeout expires, goes up again, and
so on.
The same logic apply in R5, you need to filter IGRP as well.
Also, since there is no following route-map statement, is that
taken to be an explicit deny?
>>> Without the following entry, implicit deny works here.
Now, if as another poster said, OSPF by default adds internal,
why do we have to add it here?
>>> By default OSPF redistribute internal only but we need to put external
as well. Remember there're IGRP route which need to be known by R8 and R1.
Maybe we can also use route tagging as an alternative here.
Why on R1 do we filter R8's EBGP routes and vice-versa?
>>> Because ebgp routes should be exchanged via BGP, not by redistribute
into ospf and redistribute it back again to BGP.
Does anyone have some general tips on going between OSPF<->BGP
or can explain this exercise.
>>> In this exercise, we must provide connectivity between any interface in
any router. So we need to exchange BGP routes, redistribute bgp into ospf,
and redistribute ospf (both internal and external) into bgp, and
redistribute between IGPs. The problem is we have to filter unwanted routes,
for example ospf external route from BGP doesn;t need to be redistribute
into BGP again. Beware, this lab has many potential routing loops.
Hope this will help.
Raymond.
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 08:25:45 GMT-3