Re: [v6ops] draft-moreiras-v6ops-rfc3849bis-00

["Fred Baker (fred)" <[e-mail]>]

On Aug 19, 2013, at 5:34 AM, Gert Doering <[e-mail]> wrote:

> 2001:db8 came from APNIC, that's why :-) - their delegated file lists
> 
> apnic|AU|ipv6|2001:db0::|32|20031112|allocated
> apnic|AU|ipv6|2001:dc0::|32|20030124|assigned
> 
> so an extention to /29 would technically be possible, to a /28 won't.

Hmm. Tell me about 2001:da0:: and 2001:d80::? Per ftp://ftp.apnic.net/public/stats/apnic/delegated-apnic-ipv6-assigned-latest

> apnic|AU|ipv6|2001:7fa:b::|48|20050922
> apnic|AU|ipv6|2001:7fa:c::|48|20050922
> apnic|AU|ipv6|2001:7fa:d::|48|20050922
> apnic|AU|ipv6|2001:7fa:e::|48|20050922
> apnic|ID|ipv6|2001:7fa:f::|48|20050929
> apnic|CN|ipv6|2001:7fa:10::|48|20060531
> apnic|AU|ipv6|2001:7fa:11::|48|20061031
> apnic|AU|ipv6|2001:dc0::|32|20030124
> apnic|TW|ipv6|2001:dc1::|32|20030331
> apnic|JP|ipv6|2001:dc2::|32|20030529
> apnic|JP|ipv6|2001:dc3::|32|20030619


To my small mind, that suggests 2001:d80::/26 (64 prefixes), 2001:da0::/27 (32), 2001:db0::/28 (16), or 2001:db8::/29 (8). Shorter than /26 includes 2001:dc0:: and 2001:de0::, which have been allocated. The neighborhood, however, includes 2001:db8::, which we already use. I, for one, would like to see one documentation range, at least for the global unicast address space, which is to say a prefix shorter than and including 2001:db8::/32.

http://www.apnic.net/publications/research-and-insights/ip-address-trends/apnic-resource-range#IPv6Allocation notes that 2001:DB8::/29 is reserved and by definition available.

I note that we are not discussing the recommendation per se; we are narrowing in on the length of the prefix. Unless someone disagrees, I think we have pretty much agreed that something shorter than /32 makes sense.

Here's my suggestion. The 6man chairs tell me that RFC 3489 was their work group product, so it's replacement should be. I'd suggest respinning the draft as draft-moreiras-6man-rfc3849bis (and tell [e-mail] that it replaces this one). You want to do two separate things:

a) argue for a shorter prefix in 2000::/3, and make a separate-but-analogous argument for a prefix in fc00::/8.
In those, focus on need, not want. "We designed a lab that has 2^128 different addresses in it, we obviously need the entire IPv6 address space" doesn't follow. Say what you *need* and why you *need* it. While the request was for a /20, I have not heard a cogent argument for a /26 or shorter, I heard that there was a training lab somewhere that required a /27 (32 /32s) but have not heard that the intent of the lab could not have been done with 16 /32s, and observe that a nibble boundary would suggest a /28 (16 /32s).

b) in the IANA considerations section, note:
b.1) the availability of 2001:d80::/26, 2001:da0::/27, 2001:db0::/28, or 2001:db8::/29
b.2) the suggestion of fc00:db8:?::/44, which I think we more or less agreed to in the thread
b.3) the fact that this would also affect http://www.iana.org/assignments/iana-ipv6-special-registry/iana-ipv6-special-registry.xhtml#iana-ipv6-special-registry-1