Mine was a more design based question.
>
>Lets assume that I decided not go in for CR-LDP but to go with only
>LDP. In this case, the label assignment by LDP will be either
solicited
>or unsolicited downstream. In either case, the initial route it takes
>from the source to the destionation (before LDP can initiate the label
>assignment downstream) is the default IP route i.e. SPF specified in
each
>of the routers (from source to destination). This implies that we
lost
>out on MPLS's ability to route data through a LSP. Am i making a
mistake
>in this observation?
>
>Thanks
Aamer Akhter <aakhter@cisco.com> wrote:
On Oct 18, 2004, at 8:22 AM, vishwanathan mv wrote:
> The purpose of using LDP is to automate the label distribution. But
> the basic route that the packets take initially is the default ip
> routes. Is this what we wanted?
depends on how LDP is used. as you've said, LDP is for label
distribution- how those labels are used is a different matter. For
example in the case of martini you have a case of network
ingress-egress. But if your point is that regular use of LDP relies on
regular IP hop-by-hop forwarding that is a true and was the intention.
> Haven't we lost out on our power to use MPLS (design our own routes)
> - the reason for which it was designed? I know that using CR-LDP one
> can get over this.
RSVP-TE is one way to get around the hop-by-hop model.
hope
this helps
>
> Thanks in advance.
>
> __________________________________________________
> Do You Yahoo!?
> Tired of spam? Yahoo! Mail has the best spam protection around
> http://mail.yahoo.com
>
--
Aamer Akhter / aa@cisco.com
NSITE cisco Systems
-------
The MPLS-OPS Mailing List
Subscribe/Unsubscribe: http://www.mplsrc.com/mplsops.shtml
Archive: http://www.mplsrc.com/mpls-ops_archive.shtml