Re: [sig-policy] APNIC 18 Proposal
Hi Philip,
Thank you very much for your comments, and very sorry for
ambiguousness of my sentence.
My proposal is for `existing IPv6 address holders' which have already
applied and got IPv6 address from APNIC (most of them have got /32).
I don't say /32 is too small for every LIR, nor current policy does
not allow to allocate larger space than /32.
I want to fill the gap for new applicant and existing IPv6 address
holders. Let me show an example:
A-1. LIR-A got a /32 IPv6 address space two years ago.
A-2. One years ago, they began IPv6 service trial for 300 customers.
A-3. They have a plan to begin commercial service in six months.
A-4. (I guess) LIR-A cannot expand their address space at this point
because they do not satisfy subsequent allocation requirement.
A-5. (I guess) LIR-A can apply to expand their address space only when
number of their customers reaches the subsequent allocation criteria.
While,
B-1. LIR-B will start their commercial service, and apply IPv6 address
for the first time.
B-2. They can get enough address for their needs (e.g. /24) if they can
show their concrete plan.
B-3. LIR-B can start commercial service with the address /24.
I think above situation is against LIR-A, which got IPv6 address in
the past. So, I want to make it possible to expand LIR-A's IPv6 address
space at the point A-4, if LIR-A can show their concrete plan.
| >In the past, many of the organizations had requested for the minimum
| >allocation size(/32) as an initial allocation due to the following
| >reasons:
| >
| > + Based on the idea of the "slow start" in IPv4 policy, many
| > organizations believed it would be difficult to justify all of their
| > address requirements at an initial allocation.
|
| If you want to increase the minimum allocation size, this above point
| completely contradicts the solution to your problem statement, doesn't it?
|
| > + It was difficult to estimate their needs as IPv6 network was not
| > commercially developed. Many organizations requested for address
| > space for a test service in order to kick off the commercial
| > service, not for the commercial service itself.
|
| Ok, how is it hard to work out address space needs? *All* organisations I
| have worked with have simply mapped IPv6 needs onto IPv4 customers to
| estimate their requirements. If the true aim of IPv6 is to replace IPv4,
| then this is the only logical step anyone can take. So if you have 8000
| IPv4 customers, each will get a /48 should they take on an IPv6 service;
| and you apply for a /35 to cover existing expectations. If you have 60000
| customers you apply for a /32. If you have 250000 customers you apply for a
| /30. Etc. So what's the problem? Where is the difficulty? Do ISPs not know
| how many customers they have?
Above two points are LIR's thought when they had applied IPv6 address.
For first point, when early applicant got IPv6 address, their thought
for IP address might be strongly affected by the IPv4 polices of those
days.
For second point, yes, recent IPv6 address applicant looks doing so,
and if LIRs already held IPv6 adress apply now, they'll do as you say.
| > + `PROVISIONAL IPv6 ASSIGNMENT AND ALLOCATION POLICY DOCUMENT'
| > specified the initial allocation size as /35. LIRs which received
| > allocations under this policy were only allowed an upgrade of
| > their allocations to a /32.
|
| I believe they received an upgrade to a /32 if they asked for it. If you
| look in the various RIR databases, you'll see that not every ISP has asked
| for this upgrade.
Yes, it is true, but the address size is pre-defined, not baesed on
the number of customers they had.
| >In recent days, most of the ISPs learned that /32 space is too small
| >for the real scale service deployment if they cover their existing
| >IPv4 users.
|
| So, apply for more. What's the problem?
If you mean LIRs which already have IPv6 address can apply to expand
their IPv6 address space without satisfying subsequent allocation
requirement, the only point I want to say is to state clearly this.
| >Organizations currently requesting for initial allocations can simply
| >request for a larger space as the RIRs actively emphasize to their
| >communities that they are able to request for allocations greater than
| >/32, which is already a common practice.
| >
| >However, ISPs with the default address space need to design the IPv6
| >service network within the small space untill they clear the
| >subsequent allocation requirement (HD-Ratio) for more address
| >space. This makes the real IPv6 service deployment difficult,
| >especially for large ISPs.
|
| What does "default address space" mean? Is this a /32? /32 allows for 256k
| /48 assignments - this is a *huge* network for an initial rollout. If the
| initial roll-out is going to be bigger, then isn't there a well established
| APNIC process which allows this?
I think /32 is not so *huge* especially for LIRs which already have
nation wide IPv4 infrastructure.
| >Details of your proposal:
| >
| >Existing IPv6 initial allocation address holders should be able to
| >expand their address space without satisfying subsequent allocation
| >criteria if they are able to demonstrate their concrete plan. The same
| >criteria should apply as organizations requesting for an initial
| >allocation larger than /32.
| >
| >This proposal does not intend to change the current policy but to
| >apply the current allocation practice to existing IPv6 address
| >holders.
| >
| >
| >If it is possible to expand the address space under the current
| >policy, it is desirable to be documented clearly (e.g. in the
| >guidelines document).
|
| This sounds more like a misreading or misunderstanding of existing APNIC
| policy, rather than anything approximately relevant to how IPv6 addresses
| are requested from and allocated by APNIC. IMHO.
|
| I still don't understand what the problem is from reading the proposal. It
| seems to me that you are saying simply that /32 isn't enough to deploy an
| IPv6 network today, and that there are some mysterious barriers preventing
| you from applying for more address space from APNIC. It would be really
| useful to see a very clear statement as to what these mysterious barriers
| are, as then I think it might be easier for the community to understand
| what the problem is.
|
| Looking forward to some clarification... :-)
If there are other unclear points, please let me know.
Again thank you very much for your reply.
Yours sincerely,
--
Tomohiro Fujisaki, Nippon Telegraph and Telephone Corporation