iBGP Neighborship through ISP

[Pradeep H A]

Hi

Is is possible to have an iBGP peering between routers A & B as shown below:

RTR-A(AS 100)-----------ISP MPLS(AS 200)--------------RTR-B(AS 100)

and ofcousre RTR-A & RTR-B have eBGP neighborship with respective ISP PEs and we use AS Override to remove 100 from aspath & disable synchronization?

[Julio Carvajal]

Hello Pradeep,

I think this should work, BGP allows you to create a neigborship relationship even if your peer/neigbhor is not directly connected, in this case it will need to go over the ISP but as soon as the BGP requirements are accomplished for each of them to stablish the relationship this will happen,

Regards

[Muhammad Thanveer]

Hi,

I am very new to BGP.

Dear Julio I have a question.

Thanks in Advance

We know that in BGP for making a neighbour relation ship device may not be directly connected, with in same AS.

Is the above point applicable when Same AS routers are not directly connected.I mean to say like we virtual link concept in ospf do we have in BGP.

Regards
Thanveer
"Everybody is genius. But if you judge a fish by its ability to climb a tree, it will live its whole life believing that it is a stupid."

[Pradeep H A]

Hi Thanveer,

My question was almost same as yours & Julio just answered it.

If 2 BGP routers are in the same AS and have L3 connetivity established no matter what is in between them iBGP session will get established

If 2 BGP routers are in different AS and not directly connected but have L3 connectivity established then you will have to use the command "neighbor a.b.c.d ebgp-multihop x" to establish eBGP session between them. because by default EBGP neighbors need to be directly connected.

Hope to help

Pradeep H A

[Muhammad Thanveer]

Hi Pradeep and Julio,

Thanks for that.

Pradeep, Please mark the question answered it may help others.

Regards
Thanveer
"Everybody is genius. But if you judge a fish by its ability to climb a tree, it will live its whole life believing that it is a stupid."

[milan.kulik]

Hi,

that's all correct.

But when you say "If 2 BGP routers are in the same AS and have L3 connetivity established no matter what is in between them iBGP session will get established.", that "...no matter what is in between them..." environment should not be BGP.

Otherwise you are receiving the same prefixes advertised by eBGP and iBPG on the same router with a confusing AS_PATH info :-(

With possible effects like routing loops or flapping.

Do you have some special reason for configuring such a complicated topology?

BR,

Milan

[Pradeep H A]

Hi Milan,

So does your words "Environment should not be BGP" mean my scenario of iBGP neighborship through an ISP L3 MPLS cloud will not work?

And if same routes were received on both eBGP & iBGP wont the eBGP routes be prefered over iBGP because of AD?

I can always control what routes are advertised through eBGP & iBGP by applying route-maps to neighbors right?

Thank you in advance

[Muhammad Thanveer]

Hi I have done a basic lab.

R1 as 100------------------>R2 as 200------------------->R3 as 100

Here in R1 and R2 we have ebgp neighbor relation ship and in R2 and R3 there is no bgp relation ship, we ma also initiate it. In R1 and R3 we have a static routes to reach each of them.

Then we have a ibgp relation ship between them. I have attached the Lab.

This question atleast made me to know more about bgp. I thank ever one participoated in this.

Regards
Thanveer
"Everybody is genius. But if you judge a fish by its ability to climb a tree, it will live its whole life believing that it is a stupid."

[milan.kulik · Accepted solution]

Hi Pradeep,

ad "...does your words "Environment should not be BGP" mean my scenario of iBGP neighborship through an ISP L3 MPLS cloud will not work?"

It will work, but the ISP L3 MPLS must be working first. If the ISP L3 MPLS connection fails, your iBGP connection will fail, too. So your iBGP will not bring you any additional reliability.

ad  "And if same routes were received on both eBGP & iBGP wont the eBGP routes be prefered over iBGP because of AD?"

Not necessarilly. Look to the BGP best path selection algorithm. If the same prefix comes via iBGP with a better local preference or AS_PATH length (your case probably), e.g., it will be preferred by default.

ad "I can always control what routes are advertised through eBGP & iBGP by applying route-maps to neighbors right?"

Sure you can, but wouldn't it be easier to advertise them only to eBGP and mark by AS prepending or BGP communities?

BR,

Milan