[Chapter-delegates] Discussion Paper: An analysis of the “New IP” proposal to the ITU-T
Paul Brooks
paul.brooks at internet.org.au
Thu Jul 16 07:03:00 PDT 2020
Let me take a Devils Advocate position for a moment, preaching to the converted...
On 16/07/2020 9:10 am, Dave Burstein via Chapter-delegates wrote:
>
> But what happens if air traffic control in Hamburg (on DT's network) wants to
> coordinate with ATC in Lyon, on Orange/FT. There's no way today to guarantee
> performance across networks, although the IEFT has long had solutions. The same
> problem occurs with multi-player Pokemon Go when one player is on DT and the second
> on Vodafone.
>
> To solve that problem, the telcos want to (gradually) dump TCP-IP in favor of a
> tightly controlled "deterministic network." TCP-IP has proven its ability to adapt
> and scale. I think the efficiency claims will not be realized and the cost much
> higher. In theoretical discussions in standards, the new system works perfectly. I
> live in the real world.
Dave - are we forgetting the impact of the general public user, and application
deployers, and 'the market' inertia in all this?
We - and 'they' - used to have such a deterministic network system. It was called ATM.
Before that we also had a deterministic network system - clear channel SDH/SONET
transmission services. For a while there were parallel global and national ATM
backbones and IP backbones. There were IP-over-ATM backbones which were supposed to
provide determinism and time-base slicing. There were 'ATM Switched Virtual Circuits'
that never actually switched in practice. Over time the Internet and IP became
dominant. Why?
ATM commercial services didn't die off because they were inherently technically
inferior (although they were), they died off due to lack of customer demand, which
lead to a lack of revenue. ATM gear was more expensive, services were more expensive
and inflexible, and actual 'deterministic' services even more so - the user-base voted
with their feet and wallets and ignored 'proper deterministic' expensive networks in
favour of those good-enough-in-practice more flexible cheaper systems. It didn't
matter how 'superior' it was touted, nobody wants to expend big funds building
infrastructure that has no customer demand.
>
> TCP-IP will gradually be replaced by an alternative that is designed for QoS/network
> slicing because the carriers believe they will be paid for SLAs and guaranteed
> performance. This is desired by security agencies and the carriers think they can
> sell it to others. They smell $billions and intend to push it through. Like
> everything else in telecom, the Chinese will be the first to a large build, but
> Telefonica, DT and others intend to move forward.
>
> It will take at least 3 years for this to become practical to deploy and more likely
> 5-7.
There is now an enormous installed base of IP-based applications and systems,
developers who understand how to build applications that communicate with IP,
investment in codebase that relies on sockets, ports, IP addresses and name resolution
- heck, we haven't been able to replace IPv4 with IPv6 in *two decades*, and thats
without any concerted pushback. For anything to be deployed inside a decade, it will
need to be standardised, ratified, built into routers, debugged, gather commercial
acceptance, AND have some form of translation/convergence shim invented so that three
decades of operating systems, services, browsers, chat apps etc don't need to be
thrown out the window. Whatever comes out it won't 'kill off TCP/IP' until a decade
after it is built into native Windows, Linux, IOS and Android drivers AND all the
applications people use today. Otherwise, whatever 'they' invent will end up being an
expensive set of deterministic switched-virtual-circuit layer1/layer2 network, on
which 'the market' and customers will run TCP/IP as an overlay, because nobody in
user-land will want to re-invent the application wheels, or the entire ecosystem that
it supports.
Carriers might believe they will be paid extra $$$ and $billions for SLAs and
guaranteed performance, but beliefs don't make it so - we can point to decades of
actual measured customer behaviour, from the failure of ATM to why so few customers
ever buy 'business grade' services even when they are available, to why so many people
take slower roads to bypass a highway with tolls.
One thing I know - we won't 'win' if we frame our arguments as 'defending TCP/IP and
the Internet'. The moment we talk of 'defending' something, others will wonder why
we're trying so hard to defend something if it is so obviously superior. We won't
'win' if we focus on better technicals, or more open development methods. We won't
'win' if we treat it as a adversarial 'fight'.
The best argument I've found against the 'telehealth needs guaranteed QOS networks' is
to quietly ask the person whether, even if such a network existed, they would be happy
going under the anaesthetic on the table knowing the surgeon was on the other side of
the country, given how frequently their mobile phone drops out and how often she reads
about networks being taken offline by trench-diggers - and how comfortable she thought
health insurers and courts would be with a 'QOS guarantee certificate' from the
carrier. And how comfortable the network owners will be if telehealth is the only
lonely use-case for the expensive shiny network, because everything else that doesn't
need the QOS guarantees is on a cheaper parallel competing network.
We *may* head it off, if we can highlight that all history shows that proper
determinism is uncommercial - the extra $billions they dream of isn't there - even if
they build something out of non-IP, unless it is priced at Internet commercials or
cheaper, it is likely to be an expensive, embarrassing white-elephant with no customers.
We *may* prevail if we approach it as 'helping them to avoid their embarrassment'.
Risk of embarrassing/noncommercial expensive failure is a better motivator for
no-progress and for programs being quietly abandoned than the prospect of a fight.
Best regards,
Paul.
>
> But if we think this is not a desirable long term future for the net, we have to be
> very public and make the issues clear. Olaf's paper is sensible but oblique, and our
> beliefs are invisible in places like Focus Group 2030 and the other places -
> especially 3GPP & ETSI - where decisions are being made. (I sometimes speak to it,
> but have to earn a living and haven't had much time.)
>
> Accurate information is not enough to win a political battle with literally
> $trillions of companies and their governments are on the other side.
>
> If this matters, we can and must do much more.
>
>
>
> On Wed, Jul 15, 2020 at 9:29 AM sivasubramanian muthusamy <6.internet at gmail.com
> <mailto:6.internet at gmail.com>> wrote:
>
>
>
>
> On Thu, Apr 30, 2020 at 4:46 AM Dave Burstein via Chapter-delegates
> <chapter-delegates at elists.isoc.org <mailto:chapter-delegates at elists.isoc.org>>
> wrote:
>
> Olaf
>
> We agree on the main issues here, but we need to do a much more effective
> job to win here.
>
> tl/dr unless the issues are important to you.
>
> I learned how to win policy issues through 20 years watching silver tongues
> like AT&T's legendary lobbyist Jim Cicconi. They approach the battle by
> researching carefully "what arguments will persuade the people we need to
> reach and they will believe." On telco control, which is the issue here, we
> have great pocketbook issues you've noted in your paper. Complexity and
> control drive up costs enormously. That's a great reason to stick with IP -
> it does a brilliant job adapting. Bob and Vint work is extraordinary, not
> just historic.
>
> Something like this needs support from far more people than might be
> persuaded by _anything_ we might say about governance. Very few people care
> about ITU vs ICANN or IP vs Non-IP. "Internet governance" issues are so
> obscure even tech reporters rarely get them right. No politicians understand
> this stuff and 9 out of 10 won't accept some geeks educating them. As our
> board member Pepper (bcc'd) often says, "DC doesn't understand anything that
> won't fit on a bumper sticker."
>
> We need to take a simple, clear position that non-experts will respond to.
>
>
> If non-experts are the focus of this exercise, then please show it all in
> pictures and moving images. What is European Non-IP ? What is Chinese New IP?
> And for context, what is precisely the US battle with China? How does Europe
> differ from the US? To the extent that I understand, Europeans and Americans
> wear the same clothes and speak the same English.
>
>
> * We need to stop referring to "New IP," the Chinese proposal. Instead, We
> should talk about "European Non-IP and Chinese New IP." Otherwise, our
> positions can easily be confused with the U.S. battle with China and we
> will immediately lose much of the audience we need to reach to be
> effective. Someone we both respect in a private note recently said the
> U.S. can't lead effectively here because it will be dismissed as "more
> anti-China rhetoric" He's right. We need to make clear this is more than
> a U.S. China issue. That's why I'm putting the Europeans first.
>
> Even better, we need to find a way to "frame the issue" that will
> advance our goals by getting wider support. To win this, we need to
> define the debate. Very few people care about ITU vs ICANN or IP vs
> Non-IP. "Internet governance" issues are so obscure even tech reporters
> rarely get them right.
> No politicians understand this stuff and 9 out of 10 won't accept some
> geeks educating them. As our board member Pepper (bcc'd) often says, "DC
> doesn't understand anything that won't fit on a bumper sticker."
>
> That is right. "frame the issue", yes, but frame the issue in colorful
> pictures, for at-a-glance understanding.
>
>
> * Currently, the telcos and suppliers are doing a good job convincing some
> very intelligent people QoS is crucial for things like telemedicine and
> autonomous cars. (The best telecom economist in DC, for example.) It
> also appears obvious to a non-expert that better network control would
> bring down costs. I think those are dead wrong, but people are echoing
> the arguments. How do we answer them?
>
>
> For example, I'm writing this as the "*Telcos want to take over the Internet
> and charge more.*"
>
>
> +1 (that would be an excellent way of reaching the common man and the
> non-technical policy makers.) Lynn St-Amour usually explained it by contrasting
> the way the Internet works with Cable-Telecom pricing model.
>
>
> Also, the complexity will make the system too expensive That's not quite
> good enough. I care about these issues as they affect the cost of access,
> particularly for the less affluent.
>
>
> Pricing 'Plan's ? 'bundles' ? Monthly minimum ? Connectivity or 'Air' time
> charges ? Value Added Service Charges ? Premium Services?
>
>
> * We need multiple participants at ITU and ETSI. At the first meeting of
> FG2030 in New York, there were something like 20 from China and maybe
> half a dozen from the U.S. and a similar number from Europe. After the
> New York meeting, the ratio probably got worse. Hamadoun at the 2014
> Plenipot publicly urged Kathy to send more people.
>
> *
>
>
> Incidentally, at that meeting I was the only person to raise anything
> beyond technical issues like how to deliver holograms. Brazil and India
> then backed my proposal, that one of the "use cases" for 6G in FG2030
> should be delivering systems inexpensive enough for everyone.
> ITU needs civil society and actively encouraged us. ETSI was very
> positive about ISOC in a private conversation.
>
>
>
> * We lose if this is perceived as U.S. and allies, especially if Europe is
> split. Remember, at least 65% of the Internet (by almost everyone's
> definition except AS) is not in the US and allies. The BRICS by most
> measures now have more connections than the U.S. and Western Europe
> combined. Our work on this should visibly come from a group with
> Africans and Asians prominent. Our board members Olga Cavalli
> (Argentina) and Walid Al-Saqaf (Yemen) have the right experience. India
> now has 400 million 4G connections, more than the U.S. has people. They
> would be a crucial swing. I can think of an extremely eloquent Indian
> advocate for Free Software but I think the recommendations should come
> from our Indian chapters.
>
>
> India can be a great ally, if some one can speak to India in good, clean English
> that the phone companies are trying to impress Governments everywhere with rosy
> jargon.
>
> *
>
>
>
>
>
> On Wed, Apr 29, 2020 at 11:59 AM Olaf Kolkman via Chapter-delegates
> <chapter-delegates at elists.isoc.org
> <mailto:chapter-delegates at elists.isoc.org>> wrote:
>
> Colleagues
>
> There were some requests for a public and archived space for discussion
> of this paper. We set up a list that doesn’t require ISOC membership to
> discuss this paper (and potential future discussion papers).
>
> Hence a friendly amendment to the text below:
>
> We welcome any feedback on “An analysis of the “New IP” proposal to the
> ITU-T”. Contact the authors directly using
> newIP-discussion-paper at isoc.org <mailto:newIP-discussion-paper at isoc.org>
> or post to the discussion-papers at elists.isoc.org
> <mailto:discussion-papers at elists.isoc.org> mailing list, which is public
> and archived.
>
> —Olaf
>
> On 28 Apr 2020, at 10:22, Olaf Kolkman wrote:
>
> Dear Colleagues,
>
> With the usual cross-post apologies[*].
>
> In the run up to the ITU World Telecommunication Standardization
> Assembly (WTSA-20) later this year there has been some discussion
> about a proposal called the “New IP”. It is positioned as a top-down
> architecture to solve a number of use cases that are currently been
> developed in the ITU-T’s Future Network 2030 Focus Group.
>
> The Internet Society is carefully following the developments in the
> run-up to WTSA-20. We are trying to understand if and how the New IP
> works with the Internet as we know it, if it actually solves
> problems that cannot be solved in the Internet, and, if the ITU-T is
> developing standards, where other standards development
> organizations (SDOs) have change control.
>
> In order to get a sense of the environment we commissioned a
> discussion paper, “An analysis of the ‘New IP’
> <https://www.internetsociety.org/resources/doc/2020/discussion-paper-an-analysis-of-the-new-ip-proposal-to-the-itu-t/>
> proposal to the ITU-T.” The paper helps inform us and the broader
> community whilst the public debate around these proposals shapes up.
> It also aims to inform and shape the discussion from the Internet’s
> Society’s perspective. Eventually the debate around it will inform
> our position and the potential further evolution of the discussion
> paper itself. Note that the paper documents the Internet Society’s
> emerging opinion, but does not represent a final Internet Society
> position. Instead, we intend it as a means to gather information and
> insight from our community on the topic.
>
> We welcome any feedback on “An analysis of the ‘New IP’
> <https://www.internetsociety.org/resources/doc/2020/discussion-paper-an-analysis-of-the-new-ip-proposal-to-the-itu-t/>
> via the email address NewIP-Discussion-Paper at isoc.org
> <mailto:NewIP-Discussion-Paper at isoc.org>
>
> —Olaf Kolkman
>
> [*] This mail has been sent to various relevant mailing lists and
> featured as a blog on the Internet Society website.
>
> --------------------------------------------------------------------------------------
>
> Olaf M. Kolkman Tweets as: @kolkman
> Principal - Internet Technology, Policy, and Advocacy
> Internet Society https://www.internetsociety.org
>
> --------------------------------------------------------------------------------------
>
> --------------------------------------------------------------------------------------
> Olaf M. Kolkman
> Principal - Internet Technology, Policy, and Advocacy
> Internet Society <https://www.internetsociety.org/> Tweets as: @kolkman
> <https://twitter.com/@kolkman>
>
> --------------------------------------------------------------------------------------
> _______________________________________________
> As an Internet Society Chapter Officer you are automatically subscribed
> to this list, which is regularly synchronized with the Internet Society
> Chapter Portal (AMS):
> https://admin.internetsociety.org/622619/User/Login
> View the Internet Society Code of Conduct:
> https://www.internetsociety.org/become-a-member/code-of-conduct/
>
>
>
> --
> Editor, https://Fastnet.news <http://Fastnet.news> https://wirelessone.news
> <http://wirelessone.news>
> Reply "sub" for a free subscription to Fast Net News and Wireless One
>
> _______________________________________________
> As an Internet Society Chapter Officer you are automatically subscribed
> to this list, which is regularly synchronized with the Internet Society
> Chapter Portal (AMS):
> https://admin.internetsociety.org/622619/User/Login
> View the Internet Society Code of Conduct:
> https://www.internetsociety.org/become-a-member/code-of-conduct/
>
>
>
> --
> Editor, https://Fastnet.news <http://Fastnet.news> https://wirelessone.news
> <http://wirelessone.news>
> Reply "sub" for a free subscription to Fast Net News and Wireless One
>
>
> --
> This message has been scanned for viruses and
> dangerous content by *MailScanner* <http://www.mailscanner.info/>, and is
> believed to be clean.
>
> _______________________________________________
> As an Internet Society Chapter Officer you are automatically subscribed
> to this list, which is regularly synchronized with the Internet Society Chapter Portal (AMS):
> https://admin.internetsociety.org/622619/User/Login
> View the Internet Society Code of Conduct: https://www.internetsociety.org/become-a-member/code-of-conduct/
--
Paul Brooks
Chair, Internet Australia
paul.brooks at internet.org.au
mobile +61414366605
https://www.internet.org.au
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://elists.isoc.org/mailman/private/chapter-delegates/attachments/20200717/9b58c6c2/attachment.htm>
More information about the Chapter-delegates
mailing list