Cell Relay Archive

[George Marshall <george@marshalls.org>]

Received a message indicating that perhaps the definition of an SPVC
isn't well known, so here's an explanatory followup.

Soft permanent VCC's, or just soft PVC's, (also called survivable PVC's,
but not by the PNNI spec), are simply a standard mechanism to
re-establish a PVC automatically when needed (due presumably to link or
node failure along its route).  Soft PVC's had been implemented by some
switch vendors using their own proprietary signaling protocols, as a way
of simplifying PVC setup.  The PNNI group produced the interoperable
version.

The end station doesn't (maybe can't) signal for a call, doesn't send
OAM cells or any other exotica, it just has pre-configured VPI/VCI
values to use to send & receive on.   So the switch has nailed down PVC
values on the client link, but uses PNNI signaling to the far side
terminating switch (not endpoint) to [re-]set up the connection when
needed.

See ftp://ftp.atmforum.com/pub/approved-specs/af-pnni-0055.000.pdf,
sections 4.6 and Annex C for more explanation.

George

> There's no conceptual, or theoretical, reason why soft PVC's couldn't be
> spec'd into a public networking context.  However, the creation of
> technical specs has as much to do with market forces and companies'
> self-interests (call it politics if you wish :-) as it does technology
> (well, more of the former than the latter, actually).
> 
> In 94/95, when the work on PNNI and BICI 2.0 was being done, the
> enterprise/private network switches were all using signalling (UNI 3.0,
> completed in 93), and the switch vendors were starting to get serious
> pressure from their customers to support multi-vendor networking of
> switches.  (Can't imagine why they would want want that, they didn't
> have it in the internet ;-).  In the public sector, carriers were mostly
> trialing ATM services and working on the absolute basics of handing off
> PVC's between carriers; the first announcment (nevermind actual
> deployment) of signaling in the public network was still far into the
> future.  BICI 2.0, also completed in 95, had the first reference to
> signalling.
> 
> So, when companies assigned "resources" to get on planes and go fight
> over the contents of specs they needed, there weren't many people from
> carriers and carrier equipment vendors in the PNNI and signalling
> committee rooms.  Nor were there many enterprise and LAN switch vendor
> folks hanging around the B-ICI rooms.  So, netting it out, no one (to
> exaggerate a little) in the public sector worried much about PNNI, and
> no one in the enterprise/PNNI group had much insight into the public
> network issues, or worried about them.
> 
> So, someone in the PNNI group saw a need for soft PVC's, (which was
> basically to allow a network with signaling to accomodate an end-point
> that couldn't signal) and got it pushed thru the group, but no one with
> a carrier point of view thought it important enough to send in enough
> votes to make a difference.
> 
> So, go into the next meeting and propose a new spec ...
> 
> George Marshall

--
george@marshalls.org