Re: [hrpc] new title for draft-tenoever-hrpc-political

[Tony Rutkowski <[e-mail]>]

Hi Amelia,

Rather ironically, the "lasagna" model you describe seems conceptually
identical to the OSI Model that has remained as the network and
information communication model for the past 40 years.  See
https://www.itu.int/itu-t/recommendations/index.aspx?ser=X and
especially the X.300-series, "Interworking between networks."  And, as
you conjecture, applying legal, technical, and business norms to these
layers is a challenge, but it has worked reasonably well over the past
40 years.  One of those challenges is that the layers typically have had
different bodies and norms applied to them.  The IETF, for example, is
one of maybe a dozen organizations dealing with "lasagna" layers 3-4. 
W3C deals with some of the lasagna toppings. 

An entity like 3GPP which has to produce the massive set of
specifications to operate the entire global mobile infrastructure,
pretty much deals with the all the layers as well as the smartphone user
instruments.  This occurs because they have to provide for an a critical
interoperable global infrastructure of networks and devices that support
6 billion users and innumerable "things," including roaming.  That's why
their specifications are so tightly versioned (now on Rel.15), and the
materials and activities equal to several times everyone else put
together.  It's also all rather transparent at
https://portal.etsi.org/3gpp.aspx?tbid=369&SubTB=369,649,651,653,654,656,659,396,667,373,379,380,381,382,657,843,383,375,384,385,386,387,388,825

Similarly, an entity like ITU has to produce the treaty provisions and
the radio spectrum and telecommunication norms that allow global
networks, radio systems, and satellites to actually enable all the
global interoperability.

So you can indeed narrow your focus on only the the IETF lasagna layer
3-4.  But you should be aware that rapidly emerging NFV-SDN virtual
implementations of all these layers make that focus rather meaningless
in the future.  The ancient OSI Lasagna Model is being largely
obliterated in a new virtual potpourri of slices and containers.  3GPP
is being massively consumed with NFV-SDN instantiations of 5G in Release
16.  So to quote Fleetwood Mac's "Don't Stop" theme song, you might want
to think about tomorrow here.

--tony


On 20-Aug-18 9:50 AM, Amelia Andersdotter wrote:
> On 8/18/18 11:12 PM, Andrew Sullivan wrote:
>> On Sat, Aug 18, 2018 at 03:52:45PM -0400, Tony Rutkowski wrote:
>>> 1. The scope seems to be limited to IETF standards over the past 20
>>> years.  It  would seem a stretch to apply more broadly, but you could
>>> try if a lot of changes were made, e.g., treating the scores of other
>>> standards bodies.
>> It might well be that section 5 has gradually focussed the discussion
>> toward IETF standards.  It's not clear to me that this is a good idea,
>> because (for instance), while W3C standards are only for the web and
>> not the whole Internet, an _awful lot_ of them are directly relevant
>> to the material in the draft.  Moreover, the standards-making activity
>> at W3C is similar to, without being the same as, the IETF style.  In a
>> different way, the activity of the GSMA might be directly relevant to
>> internetworking in a few areas (more on this below).  This suggests to
>> me, however, that the draft needs to spend more time on some of those
>> other standards efforts (or else, as you say, restrict its focus
>> exclusively to the IETF.  I just think it would be a much less
>> interesting result if it applied only to the IETF, because the
>> obvious response to it would be that any issues it raises would only
>> be true of the IETF and therefore one just shouldn't use the IETF for
>> those cases).
>
> The core issue of research is knowing how to limit yourself. Should an
> IRTF RG dig down into political considerations that may have an impact
> at W3C, ETSI or 3GPP?
>
> Or rather - is your criticism of the draft such that you are concerned
> that activities undertaken in other standardisation fora directly limit
> or impact the ability of the IETF or IRTF to exhibit either value
> neutrality or political considerations, causing implicitly political
> choices made by the IETF to actually not be the moral responsibility of
> the IETF as such, but rather a consequence of what peer SDOs are doing?
>
> I'm doubtful that this line of investigation will be fruitful.
>
> I was recently introduced to a metaphor for the internet as a "lasagna
> model", where each layer of communication is independent and autonomous
> from other layers of communication, and therefore each has its own
> responsibility to optimise for good privacy and security (which was the
> particular focus of that interaction). This fits, in my view, with the
> general gist of the IETF and it's already ongoing processes.
>
> The "lasagna model" is an allegory for not discarding technical
> justifications for security in one layer of the lasagna, only because
> there is a theoretical possibility that security may come to be dealt
> with in a different layer of the lasagna. When I encountered the
> "lasagna" metaphor, it also specifically was accompanied by an
> observation that many legalistic regulatory frameworks are in fact badly
> suited to "lasagna" technologies: specific regulatory frameworks
> frequently assume that responsibilities can simply be shuffled around or
> designated to any particular party somewhere in the value chain (this is
> also something I see a lot of in Tonys and Nalinis comments, when they
> are not dealing with semantic questions).
>
> Part of the thrust of my work with Niels, has been exactly taking into
> account the "lasagnesque" qualities of IETF and internetworking
> standardisation: rather than obscuring the particular layers of the IETF
> from public view, they are raised as relevant in and of themselves.
> Broadening the scope to any number of inter-relating lasagna layers
> would detract from this illustration.
>
> best regards,
>
> Amelia
>
>
>>> 2. "Internetworking" is never defined, so who knows what that means?
>> I think it is the activity of making different networks, formally
>> independent and otherwise notionally contractually unbound to one
>> another, interoperate through a common substrate protocol with as much
>> as practical of the intelligence at or near the end of the internet.
>> TCP/IP is, in this view, an internetworking protocol because it is end to
>> end yet does not depend on transitive contractual relationships
>> through the entire path in order to allow network flows.  The ARPANET
>> NCP is _not_ an internetworking protocol, and neither is something
>> like the catanet approach with smart gateways.  Similarly, several
>> "internet of things" protocols (like ZigBee and ZWave) are not really
>> internet protocols at all, but kinds of local network that require a
>> gateway which then connects to the internet (or possibly the global
>> Internet).  Therefore, …
>>
>>> 3.  X.25 data networks, SDH/SONET, SS7, mobile telephony networks are
>>> not designed for internetworking?  Really? :-)
>> …really, or anyway sort-of-really.  There is an argument to be made
>> for X.25, though I sometimes wonder whether the VC/PVC approach really
>> was internetworking given the focus on the DTE.
>>
>> SDH/SONET is something on which you happen to be able to build
>> internets (just as Ethernet is) but it is not actually an
>> internetworking technology as I am using the term.  It most certainly
>> _is_ a network technology, so if the draft wants to talk about all
>> network technologies I agree it ought to be included.  But that seems
>> to me to be a much larger topic than the draft is discussing, and I
>> think it will be hard to expand this discussion to include (e.g.)
>> ITU-T and ISO standards effectively, given that their development
>> works quite differently.
>>
>> SS7 is not, in my meaning, an internetworking protocol because phone
>> systems simply aren't internetworks.  They rely on transitive
>> contractual relationships (sometimes those contracts are encoded as
>> treaties or national law, but they're still not merely best efforts
>> the way internets are).  They are, again, important networks.  They're
>> also networks that rely on multiple networks connected together.  But
>> that interconnection does not happen as an internet does.
>>
>> Finally, some of the parts of the mobile telelphony environment are
>> internetty, and some other parts are not.  It's hard for me to speak
>> generically.  I do think that at least many parts of 5G are not
>> intended to be internet protocols, but instead to substitute the
>> catanet style of networking for internetworking.  Whether that is a
>> good idea is an open question (but, fortunately, not one for this
>> discussion).
>>
>> Best regards,
>>
>> A
>>