From: Petr Lapukhov (petrsoft@gmail.com)
Date: Wed Mar 29 2006 - 03:22:05 GMT-3
Hello Victor,
That redistribution tree idea come to me when i realized that loopless
redistribution could be achieved simply by building spanning-tree
covering routing domains. I plan to write a more detailed letter,
illustrating my ideas a little bit later.
Now, that SPT stuff is a really funny thing to do in practice labs,
when you have extremely complex scenarios. This is a way for
you to quickly achieve full reachibility.
As for the real life, scenarios are not usually that complex,
You can use some simple heuristics to fight redistribution here
1) Read scenarios' guidelinces on redistribution. Sometimes they
tell you: use only route-tagging, or use no filtering, or filter here
and here.
Often they tell you where and what to redistribute.
1) Write down prefix-lists/acls for every protocol's prefixes. That will
help you later with filtering ALOT.
3) Choose core and edge routing protocols, mark down redistribution
points on IGP diagram (I prefer to draw colored arrows, pointing
prefix-injection paths and their role - primary, backup, etc.
4) Now watch out for NATIVE via NATIVE problems. You usually have
them, when you have two or more redistribution points between
different routing protocols. And that problmes happen on redistribution
points - i.e. on routers, that have routing information from two or more
routing domains. Just note AD of protocols on suspicious routers,
Now, as for NATIVE via NATIVE rule. That sort of problem
2006/3/29, Victor Cappuccio <cvictor@protokolgroup.com>:
>
> Hello All
>
> Imagine the situation that you know that a problem occurred into your
> redistribution in the middle of the lab (time goes by
> incredible fast) and one off your goals is to archive that native
> prefixes should be reachable through native protocol.
>
> Lets that the router that is performing the route redistribution, we
> erase the redistribution configuration. (Please sorry my english)
>
> The show ip route ROUTESOURCE should be sufficient to know which routes
> are Native from others, or there is another
> better option ? Also I saw recently an email from Petr (hey bro you look
> in good shape for the Lab) this:
>
> draw detailed redistribution diagram, place routing protocols as graph
> vertexes, with edges (routers that do redistribution)
> connecting them.
>
> e.g.
> __________(R1)_________
> / \
> / ____(R2)_____ \
> / / \ \
> OSPF--(R1)--EIGRP----(R3)----RIP
> \ /
> \___(R3)__/
>
>
> Now a quick idea: If you need a loopless routing, you should build SPT
> here.
> Choose a core protocol, and build a spanning-tree, rooted in that
> protocol.
> if you choose OSPF you may end up with
>
> _________(R1)__________
> / \
> / \
> / \
> OSPF--(R1)--EIGRP RIP
>
>
> Could anyone please clarify for me this a little bit better?? lets say
> that I have something like Lab 2 from
> InternetworkExperts
> that is
>
>
> Protocols................EIGRP(2 to 4).......
> Protocols....OSPF.(1 to 4)...................
> Routers............1.....2......3........4...
> Protocols.........RIP...
> Routers...........Sw1...
>
> Lets say that you have to perform route redistribution in all routers
> and also that native prefixes should be reachable through native protocol.
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Sat Apr 01 2006 - 10:07:40 GMT-3