BGP in Transit As

nikhil.engineer · ‎05-02-2007

Hi all,

There is R1-R2-R3 in AS 100 R1 is peering with R4 which is in AS 200 and R3 is peering with R5 in AS 300. R2 is not running BGP on it. For R3 to reach the routes on R4 we need to redistribute BGP in IGP on R1. But the redistribution is not recommended. Is there any other solution?

Rgds,

Nik

mohammedmahmoud · ‎05-02-2007

Hi,

Why don't you run iBGP between R1 and R3, and use next-hop self between both.

HTH, please rate if it does help,

Mohammed Mahmoud.

nikhil.engineer · ‎05-02-2007

R1 and R3 are not directly connected and even if they are running ibgp and if i use next-hop-self the routes wont b reachable

mohammedmahmoud · ‎05-02-2007

Hi,

They don't have to be directly connected as long as they have IP reachable (you can create a tunnel between both), and the iBGP and the next-hop self shall work, as the next-hop will be reachable via whatever the IGP is.

HTH, please rate if it does help,

Mohammed Mahmoud.

nikhil.engineer · ‎05-02-2007

Hi,

For R3 ip reachability is there till R1 but as R2 is not running BGP so the routes learned from R4 i.e. from AS 200 will not be visible to him. so how will he froward the packet?

milan.kulik · ‎05-02-2007

Hi,

just an idea:

If R2 does not need to communicate to both AS 200 and 300, what about some tunnel between R1 and R3?

BR,

Milan

Harold Ritter · ‎05-02-2007

Nikhil,

One option would be to run MPLS on these routers. This way, r3 would be able to reach destinations advertised by R1 (and vice versa) even though R2 is not running BGP and therefor doesn't have global connectivity.

This is because the forwarding in the core would be done solely based on the LSP (Label Switch Path) between the two edge routers. This LSP would be build based on the loopback address of R1 (and R3 in the other direction).

Hope this helps,

Harold Ritter
Sr Technical Leader
CCIE 4168 (R&S, SP)
harold@cisco.com
México móvil: +52 1 55 8312 4915
Cisco México
Paseo de la Reforma 222
Piso 19
Cuauhtémoc, Juárez
Ciudad de México, 06600
México

mohammedmahmoud · ‎05-02-2007

Hi,

Nice solution Harold, we are working with this solution in our network, but i think that if he does't want to enable MPLS in his network he can simply build a tunnel between R1 and R3, it shall do it the same way as the LSP did it.

HTH,

Mohammed Mahmoud.

nikhil.engineer · ‎05-02-2007

Hi

Thanks Hritter, this could be one of the solution but, I dont want to do MPLS.

I have attached the configuration.

mohammedmahmoud · ‎05-02-2007

Hi,

You can simply do a GRE tunnel between R1 and R3 and configure the iBGP with next-hop self over it, this is the simplest solution for your case.

HTH,

Mohammed Mahmoud.

Harold Ritter · ‎05-02-2007

Nikhil,

As Mohammed and Milan already pointed out the options you have are either running BGP on R2 or configuring a GRE (or L2TPv3) tunnel between R1 and R3 (and vice versa).

Hope this helps,

Harold Ritter
Sr Technical Leader
CCIE 4168 (R&S, SP)
harold@cisco.com
México móvil: +52 1 55 8312 4915
Cisco México
Paseo de la Reforma 222
Piso 19
Cuauhtémoc, Juárez
Ciudad de México, 06600
México

Danilo Dy · ‎05-02-2007

Hi,

I agree with Mohammed, I have use that solution in one implementation - works well. IP GRE Tunnel between R1 and R3 and iBGP between them (R2 is in-between R1 and R3). To add I also run BGP in R2 but as a reflector client of R1 and R3.

I give him 4 points :)

Dandy

mohammedmahmoud · ‎05-02-2007

Thanks Dandy :)

Danilo Dy · ‎05-02-2007

No problem Bro :) I saw this post earlier when you were the first responder, I just forgot my password :)