[ih] IPv8...

[Jack Haverty]

That kind of situation was a major reason behind the use of "military 
scenarios" that I mentioned in another thread on this list.   Such 
scenarios were used during the circa 1980 redesign to evolve TCP 2 into 
TCP/IP 4.  We wanted to make sure that new features were likely 
approaches to solving problems surfaced in the scenarios, to be 
subsequently verified by experimental use as part of the ongoing 
research needed to achieve "rough consensus and running code".  There 
were lots of ideas tossed around, but if no one could see how some idea 
might be used, it was discarded, to avoid implementing solutions in 
search of a problem.

Somewhere in the timeline of Internet History, the notion of scenarios 
as drivers of technical choices must have disappeared.  I'm not sure 
when that happened -- perhaps when Vint left ARPA and NSF became 
involved so that the military focus faded away.  Still, there could have 
been non-military scenarios to reflect the broader scope of ongoing 
Internet development.  I don't remember any other than the problem of 
"not enough address bits".   Were there others...?

/Jack Haverty


On 4/18/26 13:02, Craig Partridge via Internet-history wrote:
> On Sat, Apr 18, 2026 at 1:52 PM John Levine via Internet-history <
> internet-history at elists.isoc.org> wrote:
>
>> The new approach to fragmentation doesn't work with anycast, as Geoff
>> Huston has often
>> noted.  Dunno whether they could reasonably have forseen that, but it's
>> still a problem for
>> large DNS systems.
>>
> As one of the co-inventors of anycast, I can cheerfully say that as IPv6
> was being developed, *nobody* fully understood anycast.  Steve Deering
> (central to IPv6) felt strongly that multicast (which he'd played a central
> role in making viable) was a better solution to all the problems that
> anycast was believed to able to solve.  But beyond that viewpoint, those of
> us who thought anycast might be useful were in the early stages of figuring
> out what it was good for, and had only a vague inkling of how it might
> behave at scale.  I still remember, initially, the sense was "oh, we
> discovered a fourth type of addressing [beyond unicast, multicast and
> broadcast], isn't that intriguing?"
>
> Craig
>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: OpenPGP_signature.asc
Type: application/pgp-signature
Size: 665 bytes
Desc: OpenPGP digital signature
URL: <http://elists.isoc.org/pipermail/internet-history/attachments/20260418/af482f47/attachment.asc>