Hey, I hope I am not wrong but the question that you are trying to ask is " when we use next-hop-self when one of our router is connected to EBGP peer then why dont we use next-hop-self when using all IBGP peers. Suppose you have 4 routes connected in a square topolgy & you are using OSPF as internal routing protocol + you have configured all the routers as IBGP peers.Ospf in this is case is just being used as the transport protocol to deliver the packets to BGP peers.Now according to this topology it is quite obvious that all 4 routers will have connectivity to each other because they are advertised in OSPF as well.
Now when a loopback on R1 is advertised in BGP it is seen on R2 & R4 with correct next hop but in case of R3 (diagonally opposite) the next-hop still remains of R1 & the best part is that we dont need to do change the next hop. The reason behind this is that R1 will have connectivity to all the routers inside so you dont have to think about changing the next hop to R1 in case of R3.What i am trying to say is when R3 can reach R1 via OSPF u dont have to change next hop on neighboring routers & of course this stands true even if you have LOGG SYNCHRONOUS rule turned ON. Now lets take an example of EBGP peers to make it a bit easier.R1 , R2 , R3 connected in straight topology with r1 & r2 being in AS 1 and r3 being in AS 2.Now in this case when r3 which is EBGP relation with r2 advertises a loopback 3.3.3.3 to R2.R2 will show the correct next hop & while advertising it to R1 it will not change the next hop.Now in this case we need to do next-hop-self on r2 & the reason behind it is that R1 has no idea about the link between R2 & R3 as it is ebgp relation & ebgp relation in this case being with ISP can never run a dynamic routing protocol.So r3 needs to know the next-hop i , e r2. I hope I answered your question & sorry for making it a big one. Regards, Omer Wali _______________________________________________ For more information regarding industry leading CCIE Lab training, please visit www.ipexpert.com
