Terence,
Thanks for your reply.
Basically, my concerns can be addressed by replacing 'reasonable justification'
with condition (a)(i) &(ii).
Thanks for this - I think I now have a better understanding of your
perspectives.
As stated at the start of section 4, the proposal is intended to replace
only section 5.9.1 of APNIC's IPv6 addressing policy - so in the format
you stated, that would be:
5.9.1. a(i) OR a(ii), OR
5.9.2. IXP OR
5.9.3. Critical Infrastructure
That is, the criteria for allocations to Internet Exchange Providers or
Critical (Internet) Infrastructure would remain unchanged. (BTW,
everyone should note that the intent of a(ii) - providing portable space
for critical civic services - is *not* the same as 5.9.3, which concerns
allocations of portable address space to Internet network infrastructure
like root domain names, etc.)
So if the change were made to the proposal as written in the last email,
this would mean multihoming by itself would *not* be an automatic
justification for a portable assignment. My mild arguments for this
would be along the same concerns as yours about reducing global routing
table impacts, since I wouldn't want to see a proliferation of routes
and AS paths just because it became a matter of status or kudos amongst
technophiles to obtain portable addressing and a 4-byte AS number simply
because their small home networks are connected to two providers, when
in principle multihoming with IPv6 could be achieved using multiple
addressing assigned from the different providers.
But I fully admit that I'm mainly trying to stimulate some discussion
from the list for my own interest, and I suspect that there will
actually be a good reason for a additional criterion of multihoming to
be included; I'd just like someone else to articulate that necessity
clearly!
I do remain concerned about the potential unintended impacts of removing
the simple "reasonable technical justification". As a hypothetical
example (of an admittedly contrived scenario), if someone wanted to
develop new routing protocols (e.g. a new BGP implementation) and wanted
a small test network which was portably addressed to do so, then I don't
think that under a strict interpretation of (a)(i) and (a)(ii) alone
that they would be eligible to do so - what would their options be then?
So I'd like to ask the list for their comments on the following:
(A) Would the members of the list prefer to:
(I) leave the proposal with "reasonable technical justification"
as the only criterion for portable allocation (i.e. for interpretation
by the Secretariat), or;
(II) change the proposal to specify explicit criteria, as
discussed between Terence and myself?
And if the proposal were changed to specify explicit criteria (instead
of "reasonable technical justification"), then:
(B) Would anyone who had previously supported the proposal then
change their position to opposing it?
(C) Should multihoming be added as one of those explicit criteria?
That is, should multihoming remain as an automatic justification for
portable address allocations?
(D) Are there any other obvious additional criteria which would be
considered essential for the proposal to be practical? (I'll note we
should be aiming for the minimum of such criteria required to make the
proposal work.)
(Please note that (B), (C) and (D) will be irrelevant if a clear
preference emerges for remaining with "reasonable technical
justification" as the criterion.)
I thank everyone for their thoughts in advance,
David
On 10/08/2012 8:49 PM, Terence Zhang YH wrote:
Hi David,
Thanks for your reply.
Basically, my concerns can be addressed by replacing 'reasonable justification'
with condition (a)(i) &(ii).
Now I am a little confused when you said 'multihoming would *not* by itself be
sufficient justification for portable assignment'
Currently portable assignment can be justified by
- Multihoming OR
- IXP OR
- Critical Infrastructure
Do you intend to change the above criteria to:
1 . a(i) OR a(ii)
2. IXP OR
3. Critical Infrastructure
or do you intend to change to:
- Multihoming OR
- IXP OR
- Critical Infrastructure OR
- a(i) OR
- a(ii)
Since the current criterias are 'OR' conditions, I assume
the latter change would be more appropriate.
Thanks & Regards
Terence
----- Original Message -----
From: "David Woodgate" dwoodgate5@gmail.com
To: "Terence Zhang YH" zhangyinghao@cnnic.cn
Cc: sig-policy@lists.apnic.net
Sent: Wednesday, August 08, 2012 10:54 PM
Subject: Re: [sig-policy] prop-101 Returned to mailing list and Newversion posted
Terence,
Thank you for stating your concerns so clearly. Would your concerns be
addressed if the term "a reasonable technical justification" were simply
deleted, and the two criteria currently identified as examples were to
become the only specific options permitted? That is, changing the
wording to:
C. Requests by organizations that have not previously received an
IPv4 portable assignment will need to be accompanied by:
(a) EITHER:
(i) Demonstration that the relevant network is statically
addressed and of a size or complexity that would make IPv6
renumbering operationally impractical within an acceptable
business period, together with evidence that dynamic or
multiple addressing options are either not available from
the relevant ISP or are unsuitable for use by the
organization;
OR:
(ii) Demonstration that any future renumbering of the relevant
network could potentially interfere with services of a
critical medical or civic nature;
AND
(b) A detailed plan of intended usage of the proposed address block
over at least the 12 months following allocation.
I'm always concerned about *not* allowing the Secretariat scope for
interpretation of the policies, because I doubt I'd be able to think of
all of the scenarios that they would encounter. However, I *think* the
criteria (a)(i) & (ii) above would cover all the situations I'm trying
to address (but I would welcome any suggestions from the list about any
realistic situations which might not be suitably covered).
Would you consider that this change (if incorporated into a new draft of
the proposal) would address your concerns?
And to those on the list that have previously expressed support, would
you continue to support the proposal if such a change were to be made?
[BTW, it may be noted that under these two criteria that multihoming
would *not* by itself be sufficient justification for portable
assignment. My argument supporting this is that, given potential options
with IPv6 for dynamic and multiple addressing, multihoming shouldn't
necessarily mean that portable addressing will always be required or
warranted, unless the situations described by criteria (a)(i) & (ii)
applies anyway. I would be happy to discuss this further with anyone on
this list if desired.]
With best regards,
David