From: Carlos G Mendioroz (tron@huapi.ba.ar)
Date: Wed Sep 15 2004 - 13:59:23 GMT-3
Be VERY careful with next-hop-unchanged implementation of some 12.2T
versions. The command works but it does not make it to the stored
config, so a reload will make it vanish :-(
James wrote:
> On Wed, Sep 15, 2004 at 11:43:52AM -0400, Jongsoo.Kim@Intelsat.com wrote:
>
>>James
>>
>>BGP protocol will choose which next hop to use for a BGP route.
>>But still it is IGP's decision which interface to use to reach the next hop.
>>
>>Nexthop 10.90.90.1 for 2.2.2.0/29 is selected on R6 by iBGP process learned
>>from R1.
>>However, which interface R6 to use to reach 10.90.90.1 is based on IGP
>>process ( OSPF in this case).
>>So R6 will forward any packet of 2.2.2.0/29 to R4 via F/R interface.
>>
>>In the same way, R4 learned 2.2.2.0/29 from two eBGP connections from R1 and
>>R6.
>>But it chooses R6's peer address as nexthop as you said higher IP address
>>(unless bgp loadbalance is enable.)
>>To determine which interface to use to reach 10.6.6.6, R4 will check IGP and
>>forward to F/R interface going to R6.
>>
>>If you make bgp nexthop unchanged on R1 so that R4 choose 10.90.90.1 as
>>nexthop for 2.2.2.0/29, then still R4 will rely on IGP to select the proper
>>interface to forward.
>
>
> Yes, being a programmer myself writing own BGPd implementation, I am familiar
> with how BGP works to select best path ;-)
>
> I think it was more of confusion between us in previous email exchanges.
> I may have misunderstood what you were saying in the first email..
>
> Anyhow, 10.90.x.x is clearly closer to reach via R1 in R4's perspective at
> least if your setup your IGP as the book requested. IGP may reroute, but as
> you say, things like that in lab is probably rare.
>
> Thanks,
> -J
>
>
-- Carlos G Mendioroz <tron@huapi.ba.ar> LW7 EQI Argentina
This archive was generated by hypermail 2.1.4 : Fri Oct 01 2004 - 15:00:44 GMT-3