[Bloat] [Ecn-sane] can we setup a

David P. Reed dpreed at deepplum.com
Tue Mar 26 13:07:16 EDT 2019

Small comment on the closing metaphor. I was thinking in the "UNIX world".
But I think this is a much  bigger deal than that.  So the comparison is really better when you think about how macOS replaced UNIX when user experience started to matter to enable growth in the world.
So if you wanted to improve the plumbing of Unix, it would have been a much bigger deal to focus on Mach as it shaped the macOS evolution.
(I do realize that Linux is now the server platform of choice, and to a very small extent, plays a role in CE routers based on WRT. But Linux is a tiny specialized part of the Internet.)
So again, I think the core question is about the IP datagram and IP forwarding components, and if you want that to evolve to be a lot better as a "neck of the hourglass", keep your work laser-focused on this vague colossus called 5G by all the participants, even though it is currently fragmented into all kinds of warring camps at the technology level and at the applicaiton level. There will be a grand convergence.
The worst possible case is that a new "neck of the interoperability hourglass" evolves in 5G-land, and IP becomes a minor and optional part of the game.
-----Original Message-----
From: "David P. Reed" <dpreed at deepplum.com>
Sent: Tuesday, March 26, 2019 11:12am
To: "Mikael Abrahamsson" <swmike at swm.pp.se>
Cc: "ecn-sane at lists.bufferbloat.net" <ecn-sane at lists.bufferbloat.net>, "bloat" <bloat at lists.bufferbloat.net>
Subject: Re: [Ecn-sane] can we setup a

This, and the general question of how to get any change like this into the IP forwarding components of existing networks, seems to be a very important and tough question.

IETF seems to be unable to mandate anything, even when there is rough consensus and working code.

The power has shifted to customers of equipment vendors.

The business innovation for those customers is now called 5G. That's not the 3GPP standard called 5G, but a vague buzzword marketing race that pretty much wants to make the Internet slowly die.

Fronts of that are:

1. IoT. Cloud-server-based. Uses IP but doesn't care about it. A new "overlay internet" that doesn't get built by IETF at all.

2. Small cells instead of hotspots. Key is that small cells are owned by a telecom operator, and though on or near private premises, the premise owners has no authority over what traffic they transport. Business model includes your access provider owning the wireless airtime in your home or business. No more Private Branch Exchange style deals. Instead you lease a cell or mesh of cells from the cable or phone company you select as your 5G provider.

3. PWAs replace websites. Eventually, 5G operators own these "enclaves" on your phone, Chromebook, Mac, ...that are controlled by the PWA vendor. The vendors, like PC ISVs used to, are closed and proprietary distributed app makers. The data of users is held in the enclaves and the "backends". Though W3C protocols are used, including WebRTC, the way they are used by these businesses involves harvesting data about people and their behavior, kept secret in cloud vaults to which each PWA has secure access. These vaults form a privatized economy of information used to predict and control user behavior through the PWAs that are hosted on user devices.

Given that evolution of a new internetwork structure, how to get the plumbing fixed?

Simple: move the ideas to be central to 5G networking.

At the moment, 5G is very much tangled with IP. That is rapidly changing as we speak.

Also note that cablecos and cellularcos are converging to this definition of 5G. Some tech is different: 802.11 is shifting to be part of cableco' offering, while short range high speed 5G NR and mmWave are the new cellular hotspots. But both are WLANs, and 802.11x is no longer peer to peer. But the overall business goals are identical. There is a competitive war.

Most of IETF wirk is irrelevant and will be time wasted. Thus vision of 5G as replacing today's Internet is the context. Think Solaris as IETF Internet, 5G as RedHat.

----Original Message-----
From: "Mikael Abrahamsson" <swmike at swm.pp.se>
Sent: Tue, Mar 26, 2019 at 9:20 am
To: "Holland, Jake" <jholland at akamai.com>
Cc: "Holland, Jake" <jholland at akamai.com>, "ecn-sane at lists.bufferbloat.net" <ecn-sane at lists.bufferbloat.net>, "bloat" <bloat at lists.bufferbloat.net>
Subject: Re: [Ecn-sane] can we setup a "how to get this into existingnetworks" get-together in Prague coming week?

On Tue, 26 Mar 2019, Holland, Jake wrote:

> Hi Mikael,
> Any operator nibbles on making this meeting happen?

Nobody else expressed any interest in this, so I kind of dropped the idea.

Mikael Abrahamsson email: swmike at swm.pp.se
Ecn-sane mailing list
Ecn-sane at lists.bufferbloat.net

Ecn-sane mailing list
Ecn-sane at lists.bufferbloat.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.bufferbloat.net/pipermail/bloat/attachments/20190326/f83e1022/attachment-0002.html>

More information about the Bloat mailing list