"Clearly ULA-C is just "cheap address space", nothing more nothing less."
Well, yes and no. Mostly yes. It is also a readily identifiable block of
said (cheap) addresses, with some operational differences from normal
(globally routable) addresses.
I, for one, have no problem at all with ULA and ULA-C, in principle.
Private address space, for people/environments, etc. to use however they
want seems like a dandy idea.
... and will present a familiar feel to those who do similar in IPv4
w/ RFC1918 addresses.
The catch that arises, IMHO, is WRT the DNS resolvability of these addresses
(#3 & #4, below).
I am mostly neutral here:
* While I see the value of having them resolve, I understand the (real |
perceived) operational impacts thereof.
* I also see the value in not having global resolution of non-globally
reachable hosts.
(I say mostly neutral because I am leaning towards the "make them
resolvable" side of the fence, having worked in lots of IPv4 VPN situations
where the destination addresses where RFC1918 (not globally reachable) and
the (forward) resolving of hosts was a royal PITA I can see lots of benefit
from having these resolve "automagically" (FWIW, exactly the situation
mentioned in the draft) ... basically, a difference between being able to
reach the node and knowing the address(es) I should be trying to reach it at
... )
(I also see the value in this address range being obviously special (i.e. -
from a single, well-known prefix, as this greatly simplifies filtering)
I guess what I am getting at ... in my mind, it boils down to:
1) Is there a reason to not support ULA/ULA-C, sans DNS?
(Keeping in line with "tell me what I can do, and how I can do it -
not what I can't do" ... or, maybe even " Be conservative in what you send
and liberal in what you accept.")
2) For ULA-C, in order for it to work we need a registrar or some sort ...
maybe IANA, maybe the RIRs, maybe a robot.
3) Is there a major operational challenge to allowing forward DNS resolution
of ULA-C addresses?
("It's my zone, and I'll put whatever entries I want in there")
4) WRT reverse DNS ... my thought would be to drop this. If it becomes
necessary for you to reverse-lookup the addresses you are talking to over
this private network connection then maybe it is time to manage that
internally ... or revisit this issue in the future, if it becomes that
prevalent?
5) Note that this does not preclude, nor prevent, PI (micro)allocations or
any other RIR-based allocation scheme/effort ... and that specifically,
these ULA-C addresses are not meant to be globally routable (nor to be seen
as "cheap PI address space").
... just more thoughts ...
/TJ
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
