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

[Mark Smith <[e-mail]>]

We have an installed base that numbers in the billions (i.e., at least the
Android and iPhone implementations). Whether or not 64 bit IIDs really need
to be 64 bits to suit all the requirements isn't really the question
anymore. It is what are the consequences and costs of changing it now?
They'd be high costs because of the installed base that assume 64 bits.

The criticism of "tradition" or "religion" would be more valid if it was
criticising something arbitrary that hadn't made it into production
implementations for many years. Those changes could have easily been made
in the mid 1990s, but not now when every IPhone, Android, Windows and
probably all other IPv6 implementations in the hands of end-users that
assume 64 bits.

I started experimenting with the Linux implementation of IPv6 in 1998. 64
bit IIDs, SLAAC and RA PIOs from then are still the same today, so a 1998
implementation of IPv6 would still work today, nearly 20 years later.
That's how long (at least) these working conventions have been in place.

Change isn't cheap in IPv6. It's much cheaper to propose significant change
for the next version of IP (IPv10 going by IANA).

Regards,
Mark.


On 17 Aug. 2017 7:05 pm, "DY Kim" <[e-mail]> wrote:

---
DY


> On 17 Aug 2017, at 17:58, Ole Troan <[e-mail]> wrote:
>
> Have you read RFC7421?

Yes, I did. With my own bias, the doc looks quite in the interest of
defending one school of thoughts.

> What problem are you trying to solve?

Not really. I find it a bit amusing that sometimes logics seem to be
overridden under the name of ‘rough consensus’.

And ‘recommend’ is one thing, and ‘mandate’ is another.

_______________________________________________
v6ops mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/v6ops