The MPLS WG Archive

[Curtis Villamizar <curtis@fictitious.org>]

In message <FFFC48AEAA5F7447929F4F0D93FCC12D9C5DAF@zcard031.ca.nortel.com>, "Da
vid Allan" writes:
> This message is in MIME format. Since your mail reader does not understand
> this format, some or all of this message may not be legible.
> 
> ------_=_NextPart_001_01C2A6D4.B61AE56E
> Content-Type: text/plain;
> 	charset="iso-8859-1"
> 
> Hi Curtis:
> 
> > Dave,
> > 
> > Thanks for the comments.  Lets hope Kireeti, Ping, George, et al are
> > in agreement.
> 
> ditto....still have a ways to go, mind you

I agree.

> <snip>
> > > 
> > > <DAVE> The outer LSP seems to use uniform mode as 
> > described, not pipe mode
> > > in order to inherit the TTL of the inner LSP. If the outer 
> > LSP was pipe
> > > mode, then it would start with a fixed value, not copied 
> > from the inner LSP.
> > 
> > The draft doesn't actually say that.
> 
> Don't quite grok, if the outer is pipe mode, it does not copy from the inner
> when the outer is pushed.

My fault for not being clear on that.

I meant the TTL draft, draft-ietf-mpls-ttl-04, doesn't say that you
MUST put 255 in the outer TTL and in fact you are allowed to put the
TTL from the lower label into the outer label.

> > Plus, the ingress can think its Uniform Mode and the egress can think
> > its Pipe Mode.
> 
> True, IMHO TTL is a bit of a messy space.

This would be a minor error but I thought it was worth mentioning what
MPLS traceroute would do with this situation.

<big snip>
> > IMHO - as someone formerly involved in operation at a very large ISP -
> > Its almost silly to discuss diagnostics from an outside IP address
> > that isn't reachable from the inside.
> 
> Agreed, simply an artifact of overlaying disjoint networks on hardware that
> uses uniform model. However, as noted above the more interesting case is
> when it is not a disjoint network, uniform model is used, and I hit a tunnel
> entrance. Suddently all LSRs along the TE tunnel have no knowledge of the
> FEC, but the downstream mappings synch up. Like I said, there are multiple
> outcomes of a single operation and knowing that matters.

That's in a comment in the second part gorp I just sent.  I didn't
solve it, just pointed it out.

> If a TE ingress implements uniform model, the downstream mapping TLV should
> indicate this in the stack returned. If the TE ingress implements pipe
> model, then the downstream mapping TLV should only return the current level,
> else I get more errors that require lots of interpretation.

There really needs to be a mapping TLV that says "I'm sticking your
bits into an LSP and setting TTL such that you'll see the hops" and
sends the FEC of the LSP that will be visible.

<small snip>
> > IMHO - PHP was a mistake.  One that we have to live with.
> 
> Agree it was a mistake, not sure living with it forever is the answer.

We're stuck with it.  Old core routers never die.  They just get moved
to the edge (somewhere).

> > Dave - I hope we're making forward progress.  It seems to me like you
> > and I are in close agreement at least on the points here.  After I
> > send the next part (which is related but should go in section 4, not
> > here) please let me know if I missed something that should be in the
> > intro part.
> 
> I think understanding is being achieved, colleagues tell me they are
> learning a lot following the thread ;-) Time will tell if we turn this into
> progress...
> 
> cheers
> Dave

Kireeti is out there.  He just has real work to do.

Regards,

Curtis