RE: draft-bonica-tunneltrace-02

["Thomas D. Nadeau" <tnadeau@cisco.com>]

>George....you are simply wrong here *if* applied architecturally correctly.

         We need to remember one of the guiding principles of the
IETF (remember Neil, this is not the ITU): working code and rough consensus.
Both exist for most of the existing "oam" solutions. As Yakov said 
recently, "the
proof is not necessarily in the pudding, but in the eating of the pudding." I
want to eat pudding.  Achitectural correctness is bunk if it does not leads to
implementations that cannot support the architecture in such a way as to a) be
implemented in one's lifetime, and b) can be sold for a reasonable price. 
If neither
can be met, then we are all just wasting our time.

>But please explain/list examples of layer violations you are aware of.
>BTW - If anyone takes what you have written at face value then all tunnels
>should be banned, ie they are *all* layer violations (I see Shahram also
>noted exactly similar).  Sadly, from what I see you may be actually correct
>in many cases (ie relying on fault management of one layer to proxy for
>inadequacies in another layer)....but you'd need to think the consequences
>of that out very carefully and the potential implications before you take it
>further.
>
>In the meantime.....
>
>Think of a leased line (as a server...any technology) that you do not
>own.....or even a duct network (yes its a real network, bottom layer in
>fact....and sets the inherited disjoint connectiveity graph for *all* layers
>above it) that you do not own, and I am aware of Andy Reid of BT here
>telling me you lost a bet with him over virtualisation here.  A 'Tunnel' is
>a colloquialism for a (more formally defined) server layer trail (which can
>be a link connection between client layer nodes).  If you don't/can't grok
>this yet please read the text book by Reid/Sexton on Broadband Networking or
>G.805.....and tell us what is factually wrong with what is in there.  I am
>finding these terse 'I know better than you' assertions not all
>constructive....esp when they are plainly wrong, like this one (*if* done
>arch correctly of course).

         I think what might be more constructive is to set out the
current problems that you as an operator (and other operators) think need 
to be
solved and then see how the existing solutions can be used to solve these
problems. Please don't just refer us again to your OAM draft, because I think
it is clear that is a non-starter.  If there are still holes after this 
exercise, lets work together
on these problems and find efficient and solvable solutions to them. This 
may not result
in an architectural correct over-all grand reference, but it is probably 
going to get us further
than arguing over what is architecturally pure and what is not, which IMHO 
is a waste
of everyone's time.

         --Tom




>Neil
>
> > -----Original Message-----
> > From: George Swallow [mailto:swallow@cisco.com]
> > Sent: 28 February 2002 17:12
> > To: Shahram Davari
> > Cc: 'Randy Bush'; Cuevas, Enrique G, ALASO; ccamp@ops.ietf.org;
> > swallow@cisco.com
> > Subject: Re: draft-bonica-tunneltrace-02
> >
> >
> > Shahram -
> >
> > > and layer violation issues.
> >
> > Any sort of a tunnel is a layer violation in and of itself.  So if you
> > have to violate those same layers to trace them so be it.
> >
> > ...George
> >
> > ==================================================================
> > George Swallow       Cisco Systems                  (978) 497-8143
> >                      250 Apollo Drive
> >                      Chelmsford, Ma 01824
> >



------------------------------------------------------------------------
Mathematics is the supreme nostalgia of our time.