[manet] Re: [IPv6]Re: IPv6 Address for Ad Hoc Networks

["Templin (US), Fred L" <[e-mail]>]

Chris, thank you for the mention of RFC5498 which was authored by my office neighbor. Although it does offer IANA allocations for MANETs, we will want an address type that is not specific only to MANETs but more generally applicable to any IPv6 network types. But, I can certainly cite RFC5498 for completeness. My time focused on MANET got started back around 1997 when the routing protocol debates were acute. But, RFC5889 managed to move forward with a key address type (the MLA) missing, so we will just have to fix that now.

Thank you - Fred

From: Christopher Dearlove <[e-mail]>
Sent: Friday, August 02, 2024 12:39 PM
To: Templin (US), Fred L <[e-mail]>
Cc: Kyle Rose <[e-mail]>; Bob Hinden <[e-mail]>; IPv6 List <[e-mail]>; [e-mail] List <[e-mail]>
Subject: Re: [manet] [IPv6]Re: IPv6 Address for Ad Hoc Networks

I was going to note to Kyle that when I first got involved in ad hoc networking, now I think 25 years ago this year, we had those discussions.
I think it might be a bit late to re-open that question.

In fact RFC 2501 is now 25 years old as an RFC, it dates back earlier of course. Are there any real world cases? Maybe.

I note that Fred refers to this draft as built on RFC 5889. But there is another RFC in this space, RFC 5498, that isn’t referenced. I would expect
this draft to reference that and explain why the addresses proposed here are not those defined there, what the problem with the RFC 5498
addresses are. (Not saying there aren’t - or that there are - just saying it should be referenced and discussed.)

On that this is an update to RFC 5889, I wasn’t a contributor, just an observer. But my observation was the surprising difficulty in getting
an agreed model that MANET nodes have to be routers, because the subnet assumption of all the world is like an ethernet simply isn’t
the case. As someone who came from a radio background (at least by then) it was remarkably difficult to explain that just because A can
hear B and B can hear C, this does not mean A can hear C - and there is nothing you can do about that. OK, you can build a data link
layer than makes it look like that. At which point you’ve done all the work at L2 and what are we doing at L3? (Allowing heterogeneous
networks with multiple interfaces was the standard answer.)

As for how addresses are then used, it was definitely one of those things that held up progressing to RFC 6130 in adding cases such as
borrowing a router address from a subnet address space, and allowing use of the same address (or different addresses) on different
MANET interfaces.

As for ad hoc networks that aren’t mobile, I think those have always been assumed to be covered by this WG. In fact the first demonstration
network my employer fielded (using OLSRv1) was just that. A better name always would have been DANET, the D standing for dynamic.
But there wasn’t a French Impressionist painter of that name.


On 2 Aug 2024, at 20:01, Templin (US), Fred L <Fred.L.Templin=[e-mail]<mailto:Fred.L.Templin=[e-mail]>> wrote:

Hi Kyle,


-----Original Message-----
From: Kyle Rose <krose=[e-mail]<mailto:krose=[e-mail]>>
Sent: Friday, August 02, 2024 11:45 AM
To: Bob Hinden <[e-mail]<mailto:[e-mail]>>
Cc: Templin (US), Fred L <[e-mail]<mailto:[e-mail]>>; IPv6 List <[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]>
Subject: Re: [IPv6]Re: IPv6 Address for Ad Hoc Networks

On Fri, Aug 2, 2024 at 2:35 PM Bob Hinden <[e-mail]<mailto:[e-mail]>> wrote:

IMHO, this draft should not specify any specific prefixes.   It should all be “To be assigned”.   Including it now is a distraction to the more
important question.


I think the question that needs to be answered is:  Is there a real need for Ad Hoc IPv6 addresses?   Focusing now on the details or which
or how many prefixes it might use doesn’t address that.

It sounds like the main example use case is a mobile ad-hoc network,
e.g., of vehicles traveling on a highway at different speeds and with
no ability to pre-define a mesh or to pre-determine routes within that
network to connected resources.

Has anyone proposed concretely building such a thing with a particular
application in mind?

MANETs are an easy-to-understand special case that clearly highlight why Multilink Local
Addresses (MLAs) are necessary, but the MLA case also applies to any variety of IPv6
local area network that may not be particularly mobile. Imagine giving a networking
novice a box of arbitrary networking gear (routers, switches, cables, etc) and then ask
them to just blindly start plugging cables into ports randomly. The network should be
able to come up and have all connected nodes self-assign MLAs and be able to ping6
any other node in the arbitrary topology using multihop routes if necessary. And
all of this without any infrastructure connections to the Internet.


I'm trying to figure out whether this is WG or RG territory.

This is proposed as an update of RFC5889. When RFC5889 was published, it was
decided to work this in the Intarea because the principles apply more broadly to
any kind of Ad Hoc network and not just those that are mobile. The current
document updates RFC5889, but it is IPv6 only hence 6man looks like the
appropriate working group.

Thank you - Fred


Kyle

_______________________________________________
manet mailing list -- [e-mail]<mailto:[e-mail]>
To unsubscribe send an email to [e-mail]<mailto:[e-mail]>