The MPLS WG Archive

[Kireeti Kompella <kireeti@juniper.net>]

Hi Shahram,

On Tue, 19 Nov 2002, Shahram Davari wrote:

> I am very much troubled by a comment raised in MPLS meeting, that
> in the future the order of first requirements and then solution would
> quite often change and that hopefully it would not require any modification
> to the already approved solution.

There was no such statement made -- at least, that was not my intent.
However, since there is this confusion, let me clarify my statement.

If a requirements doc is worked on prior to, or simultaneously with,
a solutions document, then it makes absolute sense that the solutions
doc wait on the reqts doc.

In the case of LSP ping, however, a solution document has been around
for quite a while, in fact, as a *WG* document and has matured quite
a bit.  This document clearly (a) is useful (good consensus on that);
(b) meets at least some of the requirements; and (c) is extensible.

As I said at the meeting, I fully support reviewing the solutions doc
(once it's progressed) as part of the reqts doc, and adding extensions
as deemed useful or necessary, deprecating TLVs if needed.  But we do
need to move forward; and real experience would nicely complement the
requirements document.

Now that I've explained, let me say that I was very troubled by your
request: there are no reqts docs for (a) MPLS, (b) RSVP-TE, (c) LDP,
etc.  What if someone were to say, let's make them all informational
while we write the reqts docs and evaluate them?

I lie.  I wasn't at all troubled :-)

Your question really should be addressed to the ADs, or the IESG:
should every solutions doc be accompanied by a reqts doc?  If not
(and I dearly hope not), perhaps when a solutions doc becomes a
WG doc, the WG and/or chairs adjudicate whether a reqts doc is
needed.  That way, a late-breaking reqts doc doesn't stall an
in-flight soln doc.

Kireeti.