>Tnx a lot Scott. > >I'm still not there. The more I think about it, the more I get confused. I >only have more questions. >You remark about the AS# cleared up some things though. Obviously the >AS-path attribute is the only means by which BGP is able to tell where a >routing update contains a loop, that is going thru different AS systems. > >One thing you write is that "IBGP neighbors will not >propagate routes to each other as a matter of loop protection". >Further on in your example you state that "R1 will forward it's routes to >R2. R2 will forward it's routes to R3". > >So I am confused between the difference between "propagate" (not done) and >"forwarding" (done).
Propagation refers to control information, like routing data packets. Forwarding refers to data information -- rout_ed_ rather than rout_ing_. > >If a router, running EBGP on one interface and IBGP on another, has learned >an external route (EBGP out of its AS), it will forward this route to all >IBGP peers, doesn't it? >And since IBGP peers are either fully meshed or clustered, each IBGP route >will learn the EBGP external route directly, not via another IBGP peers, but >directly from the router running EBGP and IBGP, don't they? yes, unless they get it from a reflector or a hierarchy of reflectors. > >Exactly what internal routers will IBGP peers "forward" to each other >except EGBP routes? >Only IGP routes? IGP and BGP propagate independently. > >In this case, IBGP learned routes will not be propagated unless they are >learned by IGP as well, aren't they? So if each router already has learned >the route(s) by IGP, why bother with learning them from IBGP anyway? eBGP routes usually are not propagated in the IGP. But let's say the router learns a route from iBGP, but also speaks eBGP to an external router. If it doesn't learn it through iBGP, how does it know that is a route to be advertised/propagated externally, rather than a purely internal (IGP) route? > > >Sorry about hasseling you like this. I did reread the BSCN book, but it >didn't make thing clearer, because it mostly state "what is the case", >rather then "why it is the case". > > Joep, > (CCNA, CCDA) > > > > > > >BTW: The notion of BGP split horizon, I think I got this one from the BSCN >book. BGP split horizon implies that those routes that are learned via IBGP >are not propagated to other IBGP peers, meaning you will need to have a >full-mesh IBGP. Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=35736&t=35679 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
