The MPLS WG Archive

[Eric Gray <EGray@zaffire.com>]

Kireeti,

	Alternatively, A may have computed a path that 
is AB(X)GH where (X) is an AS, or an address prefix, 
including routers C, D, E and F.  When C received the 
ERO containing (X) - which it is a part of - it could 
be configured to set up an FA-LSP through the AS.  
Correct?

> -----Original Message-----
> From: Kireeti Kompella [mailto:kireeti@juniper.net]
> Sent: Tuesday, August 22, 2000 10:05 AM
> To: chmetz@cisco.com; mpls@UU.NET
> Subject: Re: Forwarding Adjacency Setup Question
> 
> 
> > "some other LSRs" (let's say A in the example below) would signal 
> > RSVP/CRLDP messages (towards H) that pass thru a "head-end" 
> LSR (C) who in 
> > turn would establish a FA-LSP to F.
> 
> So far, so good.  Note that the path the A computes is ABCDEFGH.
> 
> > A--B--C(head-end)====D====E====F--G--H
> 
> See section 4.3.2. (LSP regions) that explains how C determines
> that an FA is needed to F.
> 
> > How does the A know to send an RSVP/CRLDP message thru C if 
> the FA-LSP 
> > (C===F) does not already exist? Shouldn't the FA-LSP appear 
> first as a link 
> > in A's topology database which would imply it has already 
> been setup?
> 
> Well, as mentioned above, the path that A computed is ABCDEFGH; but
> C intercepts this, and creates an FA to F (C is the head end of this
> FA-LSP), then strips the FA from the original ERO, and sends the path
> message for the A->H LSP directly to F.
> 
> C also advertises the CF FA-LSP so that others can see it and use it.
> 
> Kireeti.
>