Re: [v6ops] I-D Action: draft-ietf-v6ops-unique-ipv6-prefix-per-host-07.txt

[DY Kim <[e-mail]>]

---
DY


> On 18 Aug 2017, at 05:48, Brian E Carpenter <[e-mail]> wrote:
> 
> I am puzzled by that remark. The authors included a range of opinions and we worked
> hard to presents facts (both facts about the published specifications and facts
> about observed behaviour of implementations).

My apology to the authors if my comments fell personal to them.

> And the opening paragraph states
> "IPv6 routing is entirely based on variable length prefixes (also known as variable
> length subnet masks), there is no basic architectural assumption that
> n has any particular fixed value.”

… and the conclusion is to stick to the ‘fixed' value of 64 bits.

> So I'm afraid that the document is actually stating something about reality,
> whether we like it or not.

So, done is done…?

SLAAC implementations faithful to the spec should have left the prefix length open, to be configurable/settable by the admin. If not, the implementations should have been declared non-conformant with the spec.

The rationale for rfc7421 might be that the deployment base is so huge (counting to billions…?) that we’d better live with the ‘reality’. The rational might be by looking into the past of 20 years.

How many more years do we expect/wish IPv6 is going to prevail? Another 20 years? Or 100 years?

One question to ourselves could be whether to make decision in the interest of the past 20 years or far more years to come. The count now might be billions but it is just about to explode by the several orders of magnitude, with emerging IoT and disruptive applications. Shall we confine the future births in a cage or shall we set them open to explore their wild dreams not imaginable by us at this point in the networking history.

Is it a compelling reason to keep those uncountable implementations interoperable with those of the 90s? How many computer systems, OSs, programs are interoperable with those of 90s?

My loved Macintosh, first-generation MacBook, and the HyperCard don’t work/interoperate with my new machines, but that’s OK. Old guys fade away… that’s about life.

If any implementations are not in conformance with the spec, that’s them who should fade away, not the ones in conformance. Bad guys should die, not the other way around.

Nothing is too late to change or bring back to right.

That would be a school of thoughts I might join.

Unfortunately, the IETF ‘rough consensus’ seems to be the other way. Then, someone as me as a minority should just step back and keep silent since it’s the rule of the game that the majority prevails. Fine. That’s also about life…