From: sheherezada@gmail.com
Date: Tue Oct 23 2007 - 18:25:54 ART
In my opinion, BGP has the lowest overhead, because you don't run yet
another routing process on the PE (and you don't have to
redistribute).
OSPF has or had some limitations as per number of routing processes on
the PE (limited at 32), with obvious implications on scalability. I
am sure that you will find a document explaining this somewhere.
With the OSPF redistribution in the MP-BGP and viceversa, you cannot
get intra-area routes across the MPLS cloud. So if you have a direct
backup link between say two VPN sites, traffic will always flow
through the backup link, because intra area routes are preferred. The
sham link helps you overcome this issue when you want traffic flowing
across the MPLS cloud instead.
HTH,
Mihai
On 10/23/07, CCIE Abreu <ccie.abreu@gmail.com> wrote:
> Hello guys,
>
> This may sound like a question for the CCIE SP group, but I think it applies
> here as well, since it's about BGP and OSPF.
>
> Can anyone explain the main reason why BGP is widely used as the protocol to
> connect remote sites that are part of a MPLS network?
> Since we may be running OSPF in each remote site, why can't we have all OSPF
> all the way, let's say, having all edge routers as part of area 0?
>
> And what's the deal with OSPF sham links? I don't get those either.
>
> Thanks.
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Fri Nov 16 2007 - 13:11:18 ART