Re: [arch-d] ETSI launches new group on Non-IP Networking addressing 5G new services

[John C Klensin <[e-mail]>]

--On Wednesday, April 8, 2020 10:28 -0700 Bob Hinden
<[e-mail]> wrote:

> Brian,
> 
>> On Apr 7, 2020, at 9:58 PM, Brian E Carpenter
>> <[e-mail]> wrote:
>> 
>> All our troubles will soon be over, apparently:
>> 
>> https://www.etsi.org/newsroom/press-releases/1749-2020-04-ets
>> i-launches-new-group-on-non-ip-networking-addressing-5g-new-s
>> ervices
> 
> Interesting.
> 
> Seems to me that an application designer has a big choice
> here.  They can design their application/service so it works
> over the Internet (with a mix of L2 technologies that includes
> 5G), or with this over 5G only with what is apparently being
> proposed here.
> 
> IMHO, this is going to be a simple choice for most.

Right.  And that suggests one constructive thing that could be
said to them (both the ETSI and ITU-T efforts, as Carsten points
out) that is not covered by Alissa's (otherwise quite good, IMO)
note.  The letter, at least as I understood it, focused on the
"above IP" ("above waist of hourglass") solutions with which the
IETF has traditionally worked.  I don't know nearly enough to
have a competent opinion about whether it would help with the
problems they are pointing at, but we have a long history of
IP-over-foo protocols that are specific to particular physical
or link layers.  I assume that, if they wanted to design a
specific IP-over-5G protocol (completely in line with _their_
layering model, which Alissa was too polite to point out), that
the IETF would be delighted to have them do that.  They might
even discover, as some other developers of IP-over-foo protocols
have, that they gain insights into the lower-layer design of 5G
itself that enable them to make improvements.

Otherwise, in addition to your observation about application
designers (with which I agree unless those designers have their
development costs unwritten by ETSI, ITU-T, or their members)
this feels a lot like "every decade or so we need to revisit
either OSI or its layer-less predecessors because we aren't good
at learning (perhaps because of those strange-looking bells
stuck over our heads)".   If so, we should probably encourage
them to charge ahead and let us know how it works out for them
:-(

Mumble.
   john