1

I am studying the basics of BGP and my book states the following:

"BGP sessions use TCP port 179. TCP allows for adjacencies that are multiple hops away"

I understand that this allows routers that are not directly connected to eachother to form adjacencies but how exactly does TCP accomplish this?

Other IGP's hellos are not forwarded anywhere else except for the local segment (they're single-hop, if I can call them like that, even if we statically configure the neighbors), but BGP achieves multi-hop adjacencies using TCP? How?

Thank you in advance for your help

2 Answers 2

3

TCP is a layer-4 protocol, and it is used across networks. You used it to connect to this site. As long as there is a layer-3 path to the destination, then TCP can reach it.

iBGP works with multihop if you use an IGP or static routes for one iBGP endpoint to be able to reach another iBGP endpoint, then because BGP uses TCP, you can form iBGP neighbors.

The use of eBGP normally precludes the use of BGP multihop because the BGP endpoints not directly connected usually do not have a route to an endpoint that is not directly connected. That may be done, but it requires cooperation between two different AS owners, and Cisco requires a command to allow it.

4
  • Thank you for the response. Would it be possible to establish multi-hop adjacencies even if TCP wasn't used with BGP?
    – Mitrixsen
    Jun 12, 2023 at 16:20
  • multi-hop works anywhere the endpoints can reach each other - either via static routes or other routing protocols. I've run eBGP-MH over the internet. (it was part of the DR plan if the network ends up partitioned. that would, indeed, be a bad day.)
    – Ricky
    Jun 12, 2023 at 16:26
  • @Mitrixsen, "even if TCP wasn't used with BGP?" That does not make sense. BGP uses TCP. There is no other transport protocol for BGP. IGPs such as OSPF and EIGRP have their own transport protocols, but BGP uses TCP.
    – Ron Maupin
    Jun 12, 2023 at 16:52
  • @Ricky, I think I explained that. The problem is if you try to run multihop eBGP with another AS that refuses to do the same. Both ends need to cooperate to get it to work.
    – Ron Maupin
    Jun 12, 2023 at 16:54
3

IGPs like OSPF and EIGRP use Hello messages to discover neighbor routers on their own without the need for configuration.

BGP in contrast, requires neighbors (peers) to be explicitly configured. So Hello messages are not required since the router is told who the peers are.

A BGP router make a TCP connection to its peer. When that is successful, all the routing information is exchanged over that connection.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.