Hi Geoff,
Geoff Huston wrote:
Hi,
The following is my idea for the initial allocation.
1) to an applicant for initial allocation "without" any
specific IPv6 deployment plan (network design), but
an applicant has their intention to provide IPv6 service
in near future
-> allocate /32
This is the current IPv6 policy as I understand it. No change is
proposed to this minimum allocation size.
No change in policy-wise, I know.
But in the current operation of allocating the initial
block is, seems like to me, that RIR gives a block to LIR
just saying "Hey, I am going to provide IPv6 service to
the current IPv4 customers in the size of x million.
but there is no concrete IPv6 network design yet."
RIR did not restrict to give /32 to that kind of applicants,
right? Or RIR require to show their IPv6 service deployment
plan to examine when RIR allocate a larger size of block
than /32?
2) to an applicant for initail allocation "with" their IPv6
service plan and network design based on their current
IPv4 customers
-> judge the minimum initial allocation size just accomodating
enough number of /48s in their planned.
But how does one calculate "just accommodating enough number
of /48s"??? The policy points to the use of the HD Ratio as the means
of calculating this quantity of "enough".
I do not think so. HD-ratio is just set to say
"you reached the number of registering /48s upto xx.
now, you are eligible to request sub-allocation
only-if you need larger address block which you
are currently allocated."
Enough address space comes from the IPv6 network design
to accomodate the IPv4 users + additional new service
they plan if existing.
LIR can figure out how many /48s necessary for that
service. If they cannot do, they can just start with
the minimum /32 until they can figure it out.
Then, if they need 200 million, allocate /26. simple.
Not giving /21 by HD-ratio table.
Later on, if they like to have addtional space, once
they progress to reach the number of registering /48s
to about 200,000 (by HD-ratio 0.8 applied) then
they can request to have an addition /26 block (to
become /25 in total).
HD-ratio is applied only in the case of sub-allocation.
Not at the initial allocation.
It will save 4-5 bits, which is enough to extend the IPv6
life-time doubled, don't you think?
Kosuke
So if you are operating
a mass market retail offering with, say, around 5 Million customers, then
the HD Ratio of 0.8 would indicate an IPv6 address allocation of a /20,
or some 268 million /48 address blocks. i.e. a utilization efficiency
of 2%. This does appear to be a extremely low efficiency number when the
intent is to "just accomodating enough number of /48s in their
planned", as you point out.
for both cases:
when the applicant need to expand the network size later,
once they register enough number of /48s for sub-allocation
criterion set by the HD-Ratio table, they can automatically
have the sub-allocation block which is the same size of
initial allocation as the current policy (i.e., the total
block would be doubled)
* in this case, HD-Ratio of 0.8 is good for LIRs.
The issue is that this 0.8 ratio raises real concerns relating to the
total
consumption rate of IPv6 addresses, and there are conceivable scenarios
that see a risk of complete consumption of IPv6 address space within
a period of some 50 - 70 years.
I won't repeat it here, but one of the referenced documents studies this
at some length and looks at the issues that may arise as a consequence
(http://www.potaroo.net/ispcol/2005-07/ipv6size.html). Its worth a read
as a background documents relating to this proposal.
The issue here is balancing short and longer term objectives within the
IPv6 address plan, and while a very liberal assignment policy certainly
meets short term desires, there are longer term consequences. These
include the increased risk of premature address exhaustion. The
consequent issue is that of the continued erosion of perceptions of
stability, robustness and sanity related to IPv6 investment by industry
players, as it is industry who will be called upon to make significant
investments in this particular technology. The larger the sums
involved the more conservative they tend to get, and they are
already very conservative these days. So if the message is
"well, we will start out this way, but if it all works out as we
anticipate
and IPv6 gets very widely deployed then we'll need to change
the address plan on the fly", then I'm afraid that we will be doing
noone a favour! We've already been there and done that with IPv4,
and the result has been pretty bad from quite a set of
perspectives.
regards,
Geoff
Save Vocea wrote:
Dear colleagues
APNIC welcomes comments, questions, and suggestions on the following
policy
proposal:
____________________
Final call for comments: [prop-031-v002]
"Proposal to amend APNIC IPv6 assignment and utilisation requirement
policy"
____________________
This is the final call for comments on policy proposal [prop-031-v002]
"Proposal to amend APNIC IPv6 assignment and utilisation requirement
policy".
Version one of this proposal was discussed at APNIC 20. The proposal to
amend the IPv6 HD ratio from 0.8 to 0.94 reached consensus and is now
documented in [prop-031-v002]. Other parts of proposal [prop-031-v001]
did not reach consensus.
Regarding the amendment of the IPv6 HD ratio, the following
consensus was
reached:
"The SIG accepted by consensus the proposed policy process,
which requires the text proposal to be sent to the mailing
list one month before the meeting, an eight week comment
period on the mailing list after the meeting, and final
endorsement from EC."
This proposal is now submitted to the sig-policy mailing list for an
eight week discussion period. At the end of that period, if consensus
appears to have been achieved, the Chair of the Policy SIG will
ask the Executive Council to endorse the proposal for implementation.
* Send all comments and questions to: <sig-policy at apnic dot net>
* Deadline for comments: 16 November 2005
_____________________________________________________________________
Proposal details
_____________________________________________________________________
Authors: Stephan Millet <stephan (a) telstra.net>
Geoff Huston <gih (a) apnic.net>
Version: 2.0
Date: 19 September 2005
Purpose
-------
To amend the APNIC IPv6 address allocation policies regarding the
definition of the threshold value for end-site allocation efficiency.
These measures, if undertaken generally by all RIRs, and assuming that
further measures are undertaken by the addressing community regarding
the general adoption of an end-site allocation size that would be
substantially smaller than the existing default value of a /48, would
increase the anticipated useful lifetime of IPv6 to encompass a period
in excess of 100 years, in which case no further allocation policy
changes would be anticipated.
Proposal
--------
To amend the IPv6 threshold end site allocation utilisation level to
that matching an HD Ratio value of 0.94.
Impact summary
--------------
End users:
There is no impact arising from this policy proposal.
ISPs and LIRs:
With the higher threshold end-site allocation efficiency level,
when based on a 0.94 HD Ratio, ISPs will need to undertake network
address plans according to this target level.
NIRs:
These proposed policy changes are not anticipated to have any
impact on
NIR operation, other than implementation of a 0.94 density metric
to replace the existing 0.80 value, as per APNIC policy.
APNIC:
APNIC will need to amend its IPv6 criteria to reflect the altered
HD ratio.
Background material
-------------------
This material is not formally part of the policy proposal. It is
included here only for informational purposes.
1. Paper: The IPv6 Address Plan
http://www.potaroo.net/ispcol/2005-07/ipv6size.html
Geoff Huston
2. Internet Draft: draft-narten-iana-rir-ipv6-considerations-00.txt
http://draft-narten-iana-rir-ipv6-considerations.potaroo.net/
Thomas Narten
3. Internet Draft: draft-narten-ipv6-3177bis-48boundary-00.txt
http://draft-narten-ipv6-3177bis-48boundary.potaroo.net/
Thomas Narten
Geoff Huston
Lea Roberts
____________________
References
____________________
Proposal details including full text of proposal, presentations, links
to relevant meeting minutes, and links to mailing list discussions are
available at:
http://www.apnic.net/docs/policy/proposals/prop-031-v002.html
--
Savenaca Vocea, Policy Development Manager, APNIC <save at apnic dot net>
http:// www.apnic.net
ph/fx +61 7 3858 3100/99
* sig-policy: APNIC SIG on resource management
policy *
_______________________________________________
sig-policy mailing list
sig-policy at lists dot apnic dot net
http://mailman.apnic.net/mailman/listinfo/sig-policy
--
**********IPv6 Internet Wonderland!************
Kosuke Ito, Master Planning and Steering Group
IPv6 Promotion Council of Japan
(Visiting Researcher, SFC Lab. KEIO University)
Tel:+81-3-5209-4588 Fax:+81-3-3255-9955
Cell:+81-90-4605-4581
mailto: kosuke[at]v6pc.jp http://www.v6pc.jp/
Lifetime e-mail: kosuke[at]stanfordalumni.org
* sig-policy: APNIC SIG on resource management
policy *
_______________________________________________
sig-policy mailing list
sig-policy at lists dot apnic dot net
http://mailman.apnic.net/mailman/listinfo/sig-policy