[gnso-rds-pdp-wg] [renamed] Key early questions

Andrew Sullivan ajs at anvilwalrusden.com
Wed May 11 19:36:12 UTC 2016

On Wed, May 11, 2016 at 03:00:42PM -0400, James Galvin wrote:
> While I have a great deal of sympathy for this point of view, I also have a
> great deal of trouble believing that an RDS is required to exist in order to
> ensure the operational stability of the Internet.
> Logically, that argument presupposes that in order to connect to the
> Internet you are required both to identify yourself and to be accessible.

No, it does not.  What it assumes is that, if you're running Internet
_infrastructure_, you have to do that.  

> 1. Enterprises routinely setup their infrastructures so that only known
> devices can connect to them.

In that case, those devices aren't really internetworking.  They're
simply consumers of services offered on the Internet, but do not
themselves act as internetworking devices.  To the extent that the
unknown devices can connect when, say, an outbound connection is
initiated, then it's not "only known devices" but "only known devices
and unknown devices we have contacted"; but once that communication is
set up, the enterprise devices _are_ internetworking.

> In addition, they also routinely fail to share
> that detailed level of contact information with the rest of the Internet.
> The enterprise contact information might itself be hidden behind a proxy or
> privacy service.

To the extent that technical contacts are so obscured in a way that
one can't reliably contact them when there's a real problem, the time
to resolution of problems with that domain is made longer.  That's the
definition of reduction of operational stability on the Internet.  
> 2. Access to the Internet is routinely provided to random unknown devices by
> all sorts of Internet cafes around the world.  The Internet functions more
> or less just fine with these devices coming and going.

But none of those people are providing Internet infrastructure that
way, so I don't think this is a counterargument.

> 3. Nation states around the world are stating that contact information for
> Internet related elements may not be shared outside the nation state.  The
> Internet functions just fine without this information being shared.

My view is that the attempts of nation-states in this area are both
futile and quite harmful to the proper functioning of the Internet.  I
see this regularly in our operations in some gepolitical regions.  The
plain fact is that in many such cases, the nation state in question
are using a persuasive definition of "Internet" that doesn't maximise

> an RDS.  If NASA can’t be contacted then NASA loses.  Comcast will have to
> deal with its customers some other way, which it ultimately did in this
> scenario and will likely do again when other circumstances require.

Well, NASA loses _and_ Comcast loses.  That was real money, in staff
time and revenue opportunity and so on, that was lost by Comcast.  The
technical reality of the Internet is that it works less well when
infrastructure operators can't be contacted.  So that's the need.  Nevertheless,

> My point is simply, from a technical point of view, if I’m willing to accept
> your help then I’ll make myself known and accessible.

I agree that there are no Protocol Police and there isn't really a way
to insist that people use the Internet it a way that tends to improve
the overall functioning of that Internet.  

> The Internet of Things is coming, or may already be here depending on your
> point of view.  Do you seriously think any other operational model is going
> to work?

Well, not if we actually get the Catanet of Things in Silos that many
vendors seem to be attempting to provide.  And if we get actually
internetworked Things that are automatically interacting with one
another and making connections independently, then it really _is_
going to be necessary to figure out how to repair those connections
automatically.  To the extent those things are real infrastructure and
not ad hoc meshes of little devices, the ability to reach operators of
the infrastructure will become more urgent, not less.

Best regards,


Andrew Sullivan
ajs at anvilwalrusden.com

More information about the gnso-rds-pdp-wg mailing list