Re: Admission field in QSPEC -- was RE: Review ofdraft-ietf-nsis-qspec-18.txt

Gerry

I read the 'L-bit' idea before I commented and I have problems with 
it.  In any of the 6 scenarios listed below, when is it appropriate 
to use L = 1? In other words, when does the reservation signaling 
protocol not believe it is the only one in use, and what difference 
would it make it is was or wasn't?

I'm suggesting that 'translation' makes the signaling at any stage of 
any scenario make everything look e2e (L=0), which makes L=1 rather 
pointless, I believe.

Now, if there was a trusted indication that informs all reservation 
protocols along a signaling path that there was no reservation 
protocol used within a part of the e2e path, that would be useful, I 
think - but that's not what you are suggesting.

James

At 12:51 PM 3/9/2008, Gerald Ash wrote:
>James,
>
>Responses below.
>
> > Gerry
> >
> > So, let me get this straight (from you), you are backing only an
> > Admission priority indicator (a field) that travels only end2end or
> > not at all (because that's the way "all SPs and vendors have agreed
> > to")?
>
>Hardly.  I don't think summaries like that help get to some 
>conclusion/consensus here.  You need to go back and read the thread 
>a little more carefully.
>
>What I am 'backing' most recently is stated in my last post 
><http://www.ietf.org/mail-archive/web/nsis/current/msg08279.html>http://www.ietf.org/mail-archive/web/nsis/current/msg08279.html 
>(in response to Martin's request to also accommodate the rsvp 
>approach in nsis) proposed a way to accommodate both the rsvp 
>approach and nsis approach by using an L bit to indicate:
>
>    L = 0 - admission priority field has end-to-end significance
>        (above admission priority values in IANA registry apply)
>    L = 1 - admission priority field has local significance only
>        (admission priority values set by local domain administrator)
>
>If both nsis and rsvp incorporate the L bit, then both approaches 
>could be accommodated.
>
> >
> > If true, you only see one of the following scenarios working
> >
> > Scenario 1 - RSVP e2e
> >        +----------+   +----------+   +---------+   +---------+
> >        |   RSVP   |<->|   RSVP   |<->|  RSVP   |<->|  RSVP   |
> >        +----------+   +----------+   +---------+   +---------+
> >          Domain1        Domain2       Domain3        Domain4
> >
> > Scenario 2 - RSVP start/end, NSIS somewhere in the middle
> >        +----------+   +----------+   +---------+   +---------+
> >        |   RSVP   |<->|   NSIS   |<->|  RSVP   |<->|  RSVP   |
> >        +----------+   +----------+   +---------+   +---------+
> >          Domain1        Domain2       Domain3        Domain4
> >
> > Scenario 3 - RSVP start/end, NSIS as the middle
> >        +----------+   +----------+   +---------+   +---------+
> >        |   RSVP   |<->|   NSIS   |<->|  NSIS   |<->|  RSVP   |
> >        +----------+   +----------+   +---------+   +---------+
> >          Domain1        Domain2       Domain3        Domain4
> >
> > Scenario 4 - NSIS start/end, RSVP as the middle
> >        +----------+   +----------+   +---------+   +---------+
> >        |   NSIS   |<->|   RSVP   |<->|  RSVP   |<->|  NSIS   |
> >        +----------+   +----------+   +---------+   +---------+
> >            Domain1        Domain2       Domain3        Domain4
> >
> > Scenario 5 - NSIS start/end, RSVP somewhere in the middle
> >        +----------+   +----------+   +---------+   +---------+
> >        |   NSIS   |<->|   NSIS   |<->|  RSVP   |<->|  NSIS   |
> >        +----------+   +----------+   +---------+   +---------+
> >          Domain1        Domain2       Domain3        Domain4
> >
> > Scenario 6 - NSIS e2e
> >        +----------+   +----------+   +---------+   +---------+
> >        |   NSIS   |<->|   NSIS   |<->|  NSIS   |<->|  NSIS   |
> >        +----------+   +----------+   +---------+   +---------+
> >            Domain1        Domain2       Domain3        Domain4
> >
> > The only one you see working is scenario 6 because it is NSIS e2e.  I
> > know this is the NSIS WG list, but do you believe it is realistic or
> > idealistic to advocate (what seems like) a
> > my-way-or-I-oppose-all-other-ways direction?
>
>Hardly.  Again, your phrasing leaves a lot to be desired here.
>
> >
> > Shouldn't there be a means for interoperability without 100%
> > comformance to 1 way of doing a thing?
> >
> > Just looking for clarification with diagrams, so I can visualize what
> > you're really saying.
>
>All the above scenarios 1 --> 6 would work if both rsvp and nsis 
>incorporated the L bit, as proposed.  Once again, that proposal is 
>to accommodate *both* approaches, as suggested by Martin.  If people 
>have alternative proposals, please go ahead and make them.
>
>Jerry
>
> >
> > James
> >
>
>
>Never miss a thing. 
><http://us.rd.yahoo.com/evt=51438/*http://www.yahoo.com/r/hs>Make 
>Yahoo your homepage.