Re: draft-ietf-tewg-restore-hierarchy-00.txt -> CCAMP OCT-12

[Jim Boyle <jimpb@nc.rr.com>]

I am familiar with a few large deployments of LSPs w/in flat IGPs.  Where
large hovers upwards to 10k.

However, do you refer to anything beyond what's specified in:

draft-ietf-mpls-rsvp-lsp-tunnel-09.txt
draft-katz-yeung-ospf-traffic-06.txt
draft-ietf-isis-traffic-04.txt

If not, then I believe the intent of the requirement was an order of
magnitude beyond what is possible today (e.g. take 10k -> 100-500k lsps)
This was presumably under the context of how one might support VPNs, edge
to edge, which required SLAs, which for some reason must equate to
tunnels.  (too much saki?)

As for doing 10k or so in one IGP, it seems to work as you note, so there
isn't really a percieved requirement for additional hierarchy at this
time, as no one wants to step up and say "I wanna deploy a million LSPs in
my network, and the current suite just doesn't cut it!"  (If this captures
your sentiment [the reader, not vijay] - speak up)  If we're fine w/ what
works today, it wouldn't be premature to not require more :)

Jim



On Wed, 3 Oct 2001, Vijay Gill wrote:

> On Wed, 3 Oct 2001, Jim Boyle wrote:
>
> >
> > Note that this draft has some changes, in particular. the hierarchy
> > requirements have been pulled back a bit.
> >
> >         Hierarchy
> >         (1) extremely large number of LSPs within flat IGPs
> >         (2)  ability to signal LSPs across IGP hierarchy (e.g. OSPF areas)
> >
> > (1) has been dropped, (2) is still there.
>
> Some current mpls TE implementations that I'm familiar with and have
> worked on, use a large (for some value of large) number of LSP's within a
> flat IGP, so dropping #1 might be a bit premature unless I'm missing
> something here (which is possible)
>
> /vijay
>
>
>
>