Transcript

APNIC 23
APNIC Fees Discussion
Thursday 1 March 2007
09:00 - 12:30

RANDY BUSH: My watch says it's nine o'clock. Welcome to the APNIC Fees WG, or something, I'm not sure what we are. We're not a SIG. Welcome to the APNIC SIG WG, Fees WG or something.

This is the agenda. We have two sessions today. I'm Randy Bush and this is Ming-Cheng Liang, Co-chair. Our guess at how timing will work out is the first session will be these formal, or semi-formal, talks, which will be some history by Che-Hoo, I'll do some statistics that I've dredged out of APNIC, Paul's going to talk about some of the trade-offs, and alternatives, etc. He will also then report on the straw poll of whoever was willing to fill out a webpage. And then, just to get a comparison from some people who have gone down a path similar to this in the past, Axel will tell us about how RIPE does these things in a very algorithmic way.

And then the second session, we're guessing this will be about break time, the second session is open microphone. So far these three talks have signed up. We'll try to keep them to a maximum of 10 minutes each for talks so we have time to argue and throw things at each other and cry and whine. So does anybody have any suggestions and desires for changes in the agenda?

If not, Che-Hoo?


CHE-HOO CHENG: Hi. Good morning. My name is Che-Hoo Cheng and I'm one of the APNIC EC members. Probably because I am in the EC for the longest time, I've been selected to present the history, but because I joined in 1998, the work before 1998 I am not responsible for.

OK, talk about history. I want to be brief in fact give you some ideas about how the existing, current membership model was established. In fact, it was an evolution instead of someone inventing the current fee structure and things.

In the very beginning, I think from 1993 to 1995, APNIC was a project. It was not really an organisation or company. There was no membership fee. The operations were actually heavily reliant on donations. At that time, of course, the Japanese Internet community provided a lot of contribution to the operations. But somehow, because there was no guarantee of the income, we started to think, "This is not a sustainable model," so there was a proposal for a new membership fee in September '95.

Finally, it was after probably a very long discussion, the fee model went into effect in June 1996 and, as you can see, we started to have a model with large members, medium members, and small members. It was largely the membership fee for large members (US$10,000) that was set at the time and it has not been changed since then.

The only one thing that's quite interesting was that there was an enterprise member category and at that time, I think it was equivalent to the portable assignment, but it was for enterprises, it was not supposed to be for ISP's.

At that time, the membership category was self-determined. If you wanted to contribute more, you could pay US$10,000 as a large member. If you didn't want to pay that much you probably would join as a small member. And there was a membership agreement. You can, if you are interested, go along the links to see the details. I read through it last night and I think it's just too much.

After November 1996, there was some slight change. I really don't know why, but the enterprise membership category was taken out, probably because the people involved then didn't think that there was a need to serve the enterprise members directly.

And gradually, because of the unique cultural issues and also the language issues in certain countries, there was a need to have a model like the NIR model right now. So, in fact, it started in about January 1997. The service to the members of these so-called confederations at that time did, in fact, give a lot of extra workload to the APNIC staff, and so there was a proposal to have a per-member fee, which we now we call per-NIR member fee. This was a fixed fee at that time to ensure APNIC has sufficient resources to continue the operations to serve the confederation requests.

So the proposal was in effect in, I think, May 1997. But somehow the confederation model at that time did allow people to set up ISP confederations serving their own group of companies instead of serving the country or economy, like the NIRs are doing, so somehow it evolved into two somewhat different sub-models. I'm not quite sure how to call this second sub-model. NIRs right now are very clear but the other model is sometimes called ISP confederations, and sometimes just called confederations. Anyway, it's a model for a group of ISPs who probably have some close relationships to get together and get services from APNIC.

In 1998, the per-address fee was introduced because the US$500 per confederation member was not easy to implement, especially in the developed countries which had NIRs. The requests actually increased and imposed a lot of workload to APNIC staff so the original model was not a good model. So there was a per-address fee so the fee was tied up with the number of addresses requested. So this became a fairer model than the previous model.

And then, maybe because APNIC was growing, and the Internet was growing very fast, there was a need to allow people to pay more, so a new category called 'very large' category was introduced. But still the membership category was self-determined.

But after Paul joined (refers to date on slide), there were a lot of issues to serve the NIRs and ISP confederations, so at that time Paul, together with the APNIC EC, then decided to suspend new NIRs and ISP confederations and, of course, I think we didn't stop them. We wanted to streamline the process with the pre-existing members first before we re-opened it again.

In March 2000, there was a big change. Before then the membership category was self-determined, but it was not a good model especially when the ISPs from Hong Kong chose to pay the lowest amount, which was not very fair to the others. So a new model was introduced. That is when the minimum membership category, or membership tier, would be determined by the amount of address space that you were holding. And it was then the three bits between membership tier model was introduced. I don't recall the details, but the three-bits were probably somewhat arbitrary.

And of course, the historical address space and experiment allocations were not counted when determining the membership fee so that the effect to the members would be minimal.

And then, at the same time, the per-address fee for IPv6 address space was introduced for NIRs and confederation members as people were starting to use IPv6.

In September 2001, there was another big change. We started to introduce three more membership categories; 'extra large', 'very small', and 'associate'. At the time, we tried to separate the membership from the allocation as we wanted to allow people to participate, to contribute in our community, without using any IP address. But when they subsequently needed to, for those people, they just needed to pay the US$625 per year membership fee for associate membership and they would not need to pay the joining fee. When they started to use IP address they needed to pay US$2,500 for the first allocation, or something like that. This was in effect in September 2001.

After almost four years, the NIR membership was reopened after the process was streamlined. I think between 1998 and 2002, the APNIC Secretariat really did a lot of work with the NIRs to streamline the process so that all the issues were resolved, especially for the NIR's. You can look at the details. The NIRs (you can only have one NIR) and another key thing is the ISPs in the corresponding economy can go to APNIC directly to get addresses. There's that option, I think that's the key idea.

RANDY BUSH: Can you form a confederation?

CHE-HOO CHENG: No, not any more.

RANDY BUSH: Not a second NIR. Can they form a confederation?

PAUL WILSON: It's still suspended.

CHE-HOO CHENG: No. Because the ISP confederation membership is still suspended.

OK, so there were no changes between 2002 and 2005. And then, because it's getting more and more difficult to change the membership fee model, due to many different issues, for requests from the big NIRs, the APNIC EC decided to give a 90% discount to the per-address fees of IPv6 address space. This was in 2005. It was the only change after this, since December 2002.

This was a brief history of how the current membership model evolved and I would like to thank the APNIC Secretariat for helping me to gather this information.

RANDY BUSH: Questions?

CHE-HOO CHENG: Or have I said anything wrong? Any questions? No? Thank you.

DEDE RUSNANDAR: Sorry, can you go back to the previous slide please?

So this is the last fee regulation from APNIC?

CHE-HOO CHENG: Last change.

DEDE RUSNANDAR: Last change?

CHE-HOO CHENG: Okay, of course, I should take away those small changes like the critical infrastructure that have small changes. I only presented the big changes. It's more related to our discussion.

DEDE RUSNANDAR: Thank you.



RANDY BUSH: Che-Hoo, thank you. To have that historical perspective was, I think, very worth while and important.

I think I can probably push that button but don't trust me. There we go. It worked!

OK. So, no more questions for Che-Hoo? OK.

I have some statistics that might help a little.

To compare financially the RIRs is interesting, okay. And I think what we see is AfriNIC and LACNIC being small developing economies, etc, I'm not so much worried about this line, their - quote – ‘profitability’. But this is their fees and the size of the expense. This is clearly one category. And now it's interesting to look at these three RIRs (APNIC ARIN and RIPE) in terms of their income and expense because it's interesting to note that APNIC has about more than half the income of ARIN and over two-thirds the expense, okay, proportionally. And, you know, a little less than one-third the income of RIPE and about half the expense of RIPE. OK? So we can think of them as kind of, you know, one, two, three, in terms of size, approximately, financially.

But what's funny is let's look at the work they do. And I'm going to measure work in IPv4 allocations. Right? See it's one of the few metrics we have. We could look at AS numbers, etc, but for the moment we're going to look at IPv4 allocations and we go historically and we can see - and the significant ones are ARIN's the blue - I don't mean to put AfriNIC and LACNIC down, it's just they are much smaller - it's that ARIN's the blue, the green is RIPE and the grey is APNIC. And you can see that ARIN kind of leads through this time period, APNIC is doing more work than everybody else in this time period and RIPE starts leading here, in the last two years, RIPE did a massive number of allocations. Right? Some suspect there's a political move in RIPE to drive everyone towards IPv6 by giving away the last IPv4 as fast as they can. But I think I actually expected that to happen here but it didn't.

But notice that these three are approximately the same amount of work that they're actually doing in each of the three regions and yet, if we remember the finances, it's very disproportionate, OK?

So let's look at the cost of the work product. So this is the revenue and expenses in 2005, OK, of the three registries - revenue is dark, expenses is medium, and allocations are yellow. So APNIC is doing about the same allocations as RIPE and ARIN, OK, but look at income and expense in comparison. So the bureaucracy costs a lot of money. And who does the most allocations in 2005? Approximately the same amount of work for all three, ballpark the same, you know? And how much money is being pushed through here? So there's clearly a difference in the service models, right?

And, in fact, we know that RIPE does a whole lot of research, software development, workshops scattered all over their community, etc, etc, which probably contribute a large amount to why income and expenses are so much larger. Their staffing model is extremely larger.

Here's APNIC's staffing model and so the finances are not really caused by a big staff build-up, because nothing's happened recently much. Very little staffing change. There was the initial one. It went up 50% there. They added a third person. So, you know. Staffing has not been - and this is a distribution of the different flavours of staff by the way if you're interested and you'll notice an administrative growth at the trade-off here and Tech Services and that gives you a feeling for how the organisation is growing and changing. OK.

So where did the money go? How come we are worried and tucking about fees? What happened? And what happened was - doesn't look like they went nuts on staffing, doesn't look like the work changed significantly. What happened? What happened was the Australian exchange rate versus the dollar. Income is in dollars. Expense -- US dollars, pardon me. Expense is in Australian dollars because of their headquarters in Australia. When the exchange rate tilts, the effective income that can be spent gets much tighter. So what we see is revenue and expense staying fairly equal but the exchange rate taking a big dip so what looks like lots of tasty revenue here turns out not being able to support much expense because of the exchange rate.

This may be the last one (referring to slides). It is.

OK, so, when we want to think of this type of issue, the question is what service model do we want? You know?

Do we want the European 'we do everything for the ISPs'? In Europe, unlike any of the other regions, the RIPE is really both the Operators Forum, it's both NANOG, APRICOT, as well as being the registry, right?

So what kind of model do we want?

And the other thing is, regarding the exchange rate, I want to know - I think you have a number of possible things you have to consider here. Either make the income in terms of Australian dollars. Make the income dynamically adjust to the exchange rate. Move to Hawaii so the expense is the same in income and expense, the currency is the same. And I'd like that of course, seeing as I come from Hawaii. And you know, but you really have to think about that. That's a little problem. And that's, well, I'm a nerd, an engineer, computer scientist, so don't listen to me, but I think there are going to be radical changes in exchange rates over the next years as the world adjusts to the United States' phoney economy, OK? So beware of this factor and make sure to have it in. OK any questions?

AKINORI MAEMURA: Thank you, Randy. My name is Akinori Maemura from JPNIC.

Could you have the slide with the comparison of the costs and allocation slides? Yes, that's it.

I'd like to add here on behalf of the NIR, that, yes, I understand and I agree that APNIC are doing great work, but I'd like to add something here. This is revenue and expense. And we have additionally, in each NIRs, we have the cost of the operations and revenue for the NIR and a portion of NIR's revenue is going to APNIC as the membership fee and, for your information, out of all APNIC allocations, maybe half of them are through NIRs, so I think this does a very, very important thing to be taken into account when we are doing this kind of analysis and discussion. That's my comment.

RANDY BUSH: Anyone else?

(Pause)

Mr Wilson.



PAUL WILSON. Good morning, everyone. I have a few things to present regarding the APNIC fee structure. Some of these, for those who have been here before, will remind you of things I have presented before. I have personally tried to move this discussion along by publishing a series of papers or a paper in a series of versions. That contribution has been an effort to try to present a practical specific way to present the fee structure issue and to break it down in a way to see and discuss the various different components of the fee structure and, in the case of my paper, a number of specific answers to each of those components or issues.

At the risk of boring you and repeating myself, I'm going to be going over some of that again today in a little bit of a different structure. Following on, I have a report back from - most of you would be aware we had an online vote, a straw poll, a non-binding vote, on a bunch of the particular issues or dimensions within the fee structure that really have to all come together in a practical way to come up with a final solution.

Let me begin. I think, in my mind, and as a response to a lot of the feedback and discussion that has been going on for quite a few years, the main issues with the fee structure have got to do with fairness and consistency. I would like to structure the next few slides to look at specific issues of fairness and consistency within the fee structure.

The first issue is the issue of fairness and equal treatment across members of all APNIC categories. What we can do, although we don't charge for IP addresses, we can look for the effective annual cost of addresses if you want to look at your membership that way. We can see big differences between small and large members. It creates a disparity in my mind, and it has been observed by others, I think, it is quite a clear disparity in the distribution of the burden in funding and also creates a barrier to entry into APNIC membership and into the axis IP addresses for smallest members.

Under the current system the lowest of fee level is arguably high and the highest is arguably low. I'll explain that later.

We have the effective annual cost of IP addresses and we can look at the cost. This graph shows on the bottom axis, it shows the total amount of address space held by a given member. Under our fee structure the total amount of address space corresponds to a certain membership tier or category and certain annual fee.

Those with more than a /10 are extra large, those are more than a /13 are very large, etc. It is a very simple sum to work out for any given number of IP addresses, what the category is and from that what the annual fee is, and from that what the effective cost for the IP addresses is. For a small member holding a /24, the annual cost is $4.88. What is more pertinent the /21, the small member has an effective cost of $1.22, but this line drops dramatically to the left. So for the extra-large member holding a /7, the annual cost for IP addresses is 0.12 of a US cent.

SPEAKER FROM THE FLOOR: That is a 1 to 1?

PAUL WILSON: The point is the address is the same. Whether you are a great member holding a lot, or a small member, you are trying to do the same business with the addresses more or less, but there is a huge disparity there. That is one issue of consistency across the treatment of APNIC members across the fee structure.

We can take this data and look at members in all, well, in five categories graphed here. I hope this is clear. But there are three bars showing the distribution, these are not absolute numbers, these are percentages relative to the contribution in terms of the number of members in each category, the amount of IPv4 address space held by the members in each category and the fees paid by each member in the category. I haven't filled the chart up with numbers, it could be too much to cope with.

What the numbers refer to is here in the small category, we have 733 members, that is some 70% of the membership on this chart is in that category. Those members pay $1.8 million per year effectively for their memberships and yet they hold 0.20 of a /8, or less than 5%, of the address base. That is, 40% of the fee income paid for less than 5% of the address base distributed.

At the other end, the five very large or extra large members hold 65% of the address space and pay $200,000 in total a year, which is less than 10% of the fees. I hope that is clear enough to illustrate the point that I'm making there.

What this corresponds to is a relatively high barrier to entry for the 733 members. It means that for new small members to join, and we hear it continually from countries like India, that it is relatively a small amount of money for a small member.

On the other hand, I think it is true to say the members in the large category are multibillion-dollar companies generally, and they have very substantial resources and yet they are funding a very small amount of APNIC's fee.

That is the first issue of fairness across the normal APNIC fee structure and the membership category. The second issue is to do with APNIC members and NIRs. The whole discussion started four years ago in the NIR regular meeting at the APNIC open policy meetings. It started with the issue of APNIC fees. Both APNIC and NIRs had quite specific problems arising. In fact, it was more than four years ago, about six years ago, it was when we were allocating at the very peak of IP address allocations in South Korea for instance, KRNIC, where the Korean NIR was being asked to pay some astronomical fees under this fee structure.

The first issue is the distribution of the funding burden, who is paying for what for the proportion of services. The problems are created by the fact we have a strange and outdated per-address fee structure which creates a big consistency issue between how an ISP served by an NIR receives and pays for its services, verses a normal APNIC member. And there is also the issue of predictability and stability of income for APNIC and the NIRs themselves.

A similar graph here, just as I explained before, three bars shows the distribution of a number of organisations. In this case, the distribution of IPv4 address holdings and the distribution of fees. But in this case the lower colour here is for conventional APNIC members. That is, it shows we have 1,385 members holding 7.53 blocks or just under 80% and paying nearly 90% of the fees that support APNIC. On the other hand, six NIRs who hold pools of addresses and 641 LIRs, that we are aware of, that hold their own direct allocations through the NIRs. You can look at the top two bars collectively, the NIRs account here for just over 30% of the number of organisations served.

The total amount of address space is nearly 60% and the fee contribution is just over 10%. That is fairly clear. I think as the NIR representatives themselves said yesterday in the NIR meeting "it is an unfair sort of situation". The point that we are all here to work on is definitely to find a way to solve this.

The per-address fee is one of the problems because under the NIR structure, and Che-Hoo Cheng explained when it came in 1997, we decided to account for NIR service costs, NIRs should pay a per-address fee, a flat per-address fee at the time of allocation, allocations received should be considered part of the NIR's address pool and contribute to the NIR's overall assessment. But there wouldn't be any ongoing address fees.

Really, the NIR is being considered as one of the extra large members paying in terms of the ongoing cost of the addresses, it is down at the level of 0.1 US cents per year, per US address, and yet the amount of services received is pretty substantial.

There are some other dimensions of fairness that I won't go into in a lot of detail about. Members in least-developed countries are quite disadvantaged. I think objectively by the fee structure and the circumstances that we have today. There are issues of consistency with non-members, that is how non-members receive and pay for services and the treatment of historical address space where they are held by members of APNIC or non-members of APNIC.

Some of these issues, well, in fact, on a separate dimension, are the issue of sustainability of the entire structure. You might not think it is an issue of fairness, but APNIC provides a whole range of needed services in the region. These services are required and in particular, they contribute to the development of the Internet in the region in a way the community and the membership does value and appreciate. I think the fairness issue there is that some of these services have much more important impact in certain parts of the region, in developing countries in particular and to the sustainability of APNIC and the sustainability of providing the services would be a pretty serious blow to a proportion, a substantial proportion of the APNIC membership.

I'll mention, as you all know, APNIC members would know, there was survey of APNIC, the APNIC stakeholder community, in terms of services and how we should allocate services in the future. It is interesting to look at the results of that which will be reported tomorrow and reflect on what many of the things that I am saying here about the important services APNIC offers, you'll find they are reflected in the views directly of the members and stakeholders who responded to the survey. I won't spoil the surprise and I'll leave it to tomorrow to expect a bit of detail there.

On the fairness issue, I wanted to highlight the developing-country issue. The United Nations defines three classes of countries which are developed, developing and least developed.

We can split the membership base into the three categories, the distribution of members, fees and IPv4 addresses. We have a similar situation here. 120 members in the least-developed countries contributing a relatively small proportion of addresses but a substantial number of addresses in terms of the value and costs in the economies and receiving a very tiny amount of IP addresses.

This is, in many people's minds, the really critical community or subgroup of countries around the world in terms of disadvantage and the need for some compensatory measures to be taken in the interests of social and economic development in the areas.

People have asked about developing countries. In our region, developing countries include India, China, Indonesia, they include the vast majority of the population and the likely development in the future. We could look at the effect, there is some disparity here, nowhere near as great, but across developing countries and look at something by way of compensatory measure there. But we are looking at a very substantial amount because of the size of the economies concerned of the overall proportion of APNIC fees and so forth.

We may not want to get into a situation where half of the APNIC membership is subsidising the other half of the APNIC membership, which will likely change in future years. It is something that can be discussed in more detail at another time I think.

I mention the issue of sustainability and fairness. Randy has already looked at this and explained this chart which shows again APNIC's sustainability is a little bit in question at the moment in terms of the annual finances.

The APNIC EC charged me, on the first day, to assure the sustainability of future services when the budget doesn't grow at all, we can sit on a consistent capital reserve.

In order to keep up with the growth of APNIC, it needs to increase and the only way the capital growth increases is to have a surplus every year. We did it in a time of rapid growth in 2000 and 2001. A big surplus of that was covered by the expensive level of the US dollar. It would have been hurting APNIC members at the time because the dollar being high, it means APNIC fees in your currencies are high. Since then, however, 2004, we have had a balanced budget which has not provided a surplus to fund the gap between our expenses from one year to the next and that is a problem.

I'll note again something here that one of the major issues that Randy mentioned is the value of the US dollar having plummeted. That has been to the serious disadvantage of the APNIC Secretariat in terms of how expenses are being expensed, the levels of expenses.

But for APNIC members, we see the lower US dollar has, depending on your own currency, had a 50% decrease in the cost of APNIC fees.

Randy, I'm probably going on too long for your proposal schedule here because I do have a few more slides here. If you would like to ding a bell and hurry me up, I'll do that. A little bit up? Okay. Thanks.

Fee structure principles, these fairness issues lead to other principles. In the proposal I have written I have tried to isolate the principles and make them as independent as possible. As you will see, the principles themselves can correspond to the parameters of the structures I have proposed. Those settings are things like the minimum fee level charged for the minimum allocation, the number of bits of difference per tier of the membership structure. Proposed discount for the members of NIRs and a couple of others, which I will explain.

That is the general idea to set the structure but to have some parameters or values that can be adjusted to set the way the fee structure will work from the time it is adopted or possibly to adjust the levels of fees without having to change the structure from year to year, or at particular points.

The first basic principle that we have embodied for some time is the annual fee calculations based on IP address holdings. It is not to say we are selling addresses. The idea here was that under the voluntary structure that Che-Hoo Cheng mentioned earlier, larger organisations were meant to contribute more. Particularly during the Asian economic crisis in the mid to late '90s, it fell apart and it was observed people weren't paying their fair share.

One way to look at it is to say the organisations with the larger numbers receive more value and have probably larger capacity to pay. The other principle here is, I mentioned before, the effective per-address cost what we want to do, one of the principles behind my proposal is to reduce the disparity across small organisations and large organisations.

There is also the idea of removing the artificial tiers so we have a continuous formula and not an arbitrary doubling of fees. The proposal was therefore to double, effectively have the fees double with every two bits of address with IPv4 space or four bits of IPv6 and also to replace the tiered structure with a continuous formula.

NIR members should have a comparable structure to normal APNIC members. This was called into question seriously yesterday in the APNIC meeting. I'm not saying it is an arguable or debatable thing, but it is a principle. The idea being an ISP in a similar situation, whether they are a member of an NIR or receiving their addresses from APNIC.

The proposal, then, in line with that principle is to apply the same type of method of fee calculation on the overall NIR itself according to the address holdings of individual members. That is not to charge the individual members of NIRs, that is the NIR's duty and responsibility but to see the contribution of the NIR collectively and also to eliminate per address fee.

Support Internet development in LDCs. We can use the United Nations definitions and look at a discount for members who are receiving address space and using address space in a set of countries designated as least-developed countries.

APNIC sustainability. Our revenue should be predictable. It is quite a challenge to have an unpredictable revenue stream when we have a staff of approximately 50. We have projects and activities which go on year after year and yet, as you will see when you have a look at financial reports tomorrow, we have really, some management to cope with, some management to be done, some challenging management to cope with this unpredictable situation with exchange rates, NIR progress fees etc.

We need to keep, as I said, maintaining a surplus if APNIC is both growing and if APNIC is going to ensure its security by holding a one-year capital reserve.

I'm just, in this case, revisiting the impacts of the proposal that was contained in version, document version 2.3.

As I mentioned, it is a parameterised system so we have a number of parameters that are straight forward. We can set the initial fee, the minimum fee for the minimum allocation or minimum address holdings we want to cope with, in this case, US $400 a year for a /24 are the two parameters. We have a doubling in IPv4 in terms of the fee contribution for every additional two bits. That is four times increase in address holdings.

An NIR discount in and an LCD discount. They are the five parameters which go into a spreadsheet and produce a whole set of figures which show what the current annual fee is for the current 1,385 members, what the total current fee is for every individual one of the members, giving us a total fee for each prefix category. We can look at the difference in fee collections in each, from the old fee and the new fee.

These things have been documented in the proposal document. These figures are updated to today, or yesterday's actual figures, in fact. This presentation will be on the website. If you want today's figures they will there. If you want more explanation, then I refer you back to the version 2.3 document.

We looked at moving on with the impact, we looked at the per-address effective fee cost, what it looked like. If we look at the new fee structure, what we have on the right-hand side is a much smaller, considerably lower fee, effective per-address cost for the smaller members than we had previously. The left-hand side is very difficult to see. We have to blow it up quite a bit to look at the differences. But you will see there is fairly minor difference at the end, but in the case of the /7, under this proposal from 0.39 cents. It is an effect of four in the fees those members will pay. It is something that has created disturbance for members in that category because fees are going up from $40,000 to $150,000 or so.


The budget impact. This spreadsheet is based on current data, it shows a 9% increase in the overall revenues of APNIC. That is not necessarily intended but that overall revenue result is what can be tuned specifically and quite easily by adjusting, for instance, the entry-level fees. It needs more work. I haven't gone into absolute detail here and to work out what the actual parameters should be because I'm talking about the principles themselves rather than the details of the values themselves.

The distribution of - under the structure I propose - the distribution of the funding burden when you are looking at members in different categories changes like this. So the third bar on the left shows exactly the value, the values that we had before but the fourth bar shows the distribution of funding burden has changed. So for the five very large members here, the funding has gone from $200,000 to $445,000. The increase only applies to probably one of those very large members, the others have a lower increase, the total of the category is about 200% increase. For the 777 small members, the total funding for APNIC members has gone from 60% to 45% or so.

RANDY BUSH: 30%.

PAUL WILSON: What we are doing is - my proposal here is to adjust the balance rather than try to implement something which might be a completely flat structure which is obviously a change of much huger proportions and debatable as to if that is actually the goal. For members, NIRs and NIR members, the distribution changes like this.

So the combined holdings of NIRs and NIR members is over 50% of the addresses being managed under APNIC at the moment. The responsibility for funding increases quite dramatically and again has caused some concern. It goes from just over 10% to just over 30% of the income for APNIC but again for 60%, corresponding to 60% of IP address-holders. That is it, I'm glad to say, you might be glad to hear for this part of the presentation and I'm just assuming there will be some discussion. It is up to Randy and maybe left to this afternoon.

RANDY BUSH: Did somebody have a specific urgent point? One.

BILLY CHEON: Could you go back to the slide that shows NIRs, IPv4 holding and general members holding and also? No, the previous one, the very first one. No, the very first slide, first presentation.

RANDY BUSH: That is where he is.

BILLY CHEON: Yes, yes, yes. Is it? NIR? Yep. Actually,

PAUL WILSON: Conventional APNIC members in the lower colours, NIRs in the bright blue.

BILLY CHEON: It seems very unfair, NIRs and NIR members are holding 50% of IP address and they pay just 20%, how unfair. But I wanted to point out there is hidden costs for NIRs. NIRs need money to build the system for IP resource management and we need to hire host masters to deal with the local ISPs so I want people to recognise this point. Yes, that is all.

PAUL WILSON: It would be interesting to do cost accounting and see how the NIR costs are and how they compare with APNIC costs and what is actually an appropriate and fair value for that discount. 50%, as you could guess, is not a scientifically derived figure. If we do a scientific analysis of where the costs are, we could still stick with the structure and look at the fair discount would be 80%.

RANDY BUSH: I think we are into the discussion that should be in the second part if we want to be on schedule.

BILLY CHEON: Okay.

RANDY BUSH: Could we continue? We shouldn't be in discussion because we are behind schedule already. If we don't.



PAUL WILSON: This won't take the 20 minutes. It is the result of the vote on the fee structure. Out of the previous proposal made, a number of questions and parameters can be drawn. Look at those separately and try to assess the members' views on each of them. After discussion with the APNIC EC it was decided to put it to an online vote. It is a formal vote in the sense APNIC members exercise and cast their votes in accordance with the voting structure, so an extra large member has 64 and an association has four, an individual one. There are 11 questions. Some are additional questions that come from the fee structure.

The problem being to make the significant changes is pretty complex.

By a principle of subdivision, we can look at the issues separately and that is the idea. The vote, as I say, is a formal vote but not a binding vote. It is quite possible when you pull all the issues out and try and make decisions about them separately you come up with a fee structure that is unworkable. It couldn't be considered to be binding at the time. We had only 76 members who voted. That is more than we generally have or often have attending an APNIC annual member meeting but still, as a proportion of the total APNIC member population, it is disappointingly low, particularly considering it is something that the online vote is available to the vast majority of members who we know use the system. About 830 votes are cast in total.

I'll give the breakdown and I'm not going to cast any opinions about them, just reporting.

The first question was under the new fee structure would we have an annual membership fee doubling with each two bits instead of each three bits of IPv4. The vote was 41% yes, 34% no.

The second question is would we use a continuous formula instead of the tiered system. 65% agreed, 16% said no.

Question three. Should we look to have an increase? 37% said effectively not, only 14% said a 5-10% increase and 9% were satisfied with a 15% increase.

Question four was about the prospect of adjusting fee levels by 5% a year. 41% said yes, 45% said no.

Question five: Under the new NIR fee structure, under the new APNIC fee structure per-address fees for NIRs would be abolished and 67% agreed.

Question six: Under the new structure, should we base NIR fees on the collective fees of all the LIRs, that is the members under the NIRs, calculated according to the same type of structure and 58% agreed with that.

Question seven: Should there be a discount, in effect recognising services provided by the NIRs? 29% of people agreed with the discount of 25% to 50% but over half were not sure about the question.

Question eight: Should NIR fees be maintained at the existing levels as a proportion of the total APNIC revenues? Regardless of how those fees are calculated and 42% of the vote said yes.

Two more - three more coming. Under the new structure, should there be a discount for members in least-developed countries? 25% people supported a 25% discount. 24% a 50% discount.

Question 10: Should historical address holdings be included in calculations and 80% said yes.

Finally, should non-members be charged the same as members and 77% said yes. Those voting were APNIC members. Non-members were not asked.

Conclusions, I have summarised the strongest results, 65% of people wanted a continuous formula. 67% wanted per-address fees abolished. 58% wanted NIR fees to be based on collective fees of their members. Nearly 50% of people supported a discount of some kind for LDCs. 80% in favour of historical resources being accounted for on the same structure and 77% asking for the same structure to apply to non-members. That is the total results. I guess there is possibly a number of things that can come up in discussion this afternoon. Thank you.

RANDY BUSH: If somebody has a point of clarification, something they didn't understand, please raise it now, discussion will be later because we are running late.

RANDY BUSH: RIPE went through this a couple of years ago and has a model that has not necessarily been suggested. There's an engineering approach - there's a model, you turn the knobs and the results come out the bottom.



AXEL PAWLIK: Or the ugly underbelly of the RIPE NCC charging scheme. I will quickly run through this. Che-Hoo has already presented us this morning with the history of the APNIC charging scheme, we have done very similar things basically.

In principle, with the RIPE NCC (that's our board, that is the RIPE NCC Services Working Group within RIPE), we do a number of surveys and talk with other people in the world also. They all give us input and with this input we draft an activity plan, a budget and also a charging scheme. All this is presented to our membership once a year prior to the general meeting and at the general meeting of course we get feedback from our members and maybe our members speak to the board and elect new members to the executive board. That's the form of decision-making.

Who does what? The number resource policies are not the RIPE NCC territory. We are just the Secretariat. The community makes the policies on which we act.

Now, the RIPE NCC activity plan and the budget on this yearly cycle is decided by the board. That's relatively new. Previously the members did that. Now we, as I said, we draft this and we publish this and of course the members use this to make their decision on a charging scheme. That of course fills the budget and then the budget is used to do all the activities that we want to do.

That's a possibility for our executive board in times of crisis, or to change the activities, something new might come up that we have to do quickly or that we don't have any money and we might have the possibility that we don't do some things any more. Now, the charging scheme, however, is the last very strong hook that the members have to tell us whether they like the activity plan, whether they like the budget or not. If they don't give us the money, we can't do anything. So here is a formal, again, also a yearly cycle, for our members to give us very strong input.

Basically, also we had contributors that were self-declaring their size and their contributions. We also had what you call confederations, we had enterprise members - that's all gone. We have also changed our voting scheme from many votes per large member and just one vote for small members to just one man, or one member: one vote. We had small, medium and large memberships. That has changed.

The main point that Randy wants me to bring forward here is how we calculate the - what is the algorithm that gives us some parameters at the top, and gives us the charging scheme, the fees, on the bottom end.

We currently have five membership classes from very small to very large and we decided at some point in time that it would be easier to just decide how many per cent of large members and how many per cent of very large members and very small members we want to have. Previous slides said there used to be a magical gap - two magical gaps between the small and the medium and the large members. That is a story that I hear from Daniel. Since I'm at the RIPE NCC, that's in 1999,that never was clear any more so we decided to get rid of that and do this one. We decided to have 1% of very large members and 3% of very small members and used an algorithm to sort members into relative size to each other.

RANDY BUSH: How do you measure size? What's the metric of size?

AXEL PAWLIK: I'll show you.

Basically that takes into account the resources that you have been allocated over time. The idea here is that it should be fair in some sense, it should reflect the amount of work that the RIPE NCC spends on your resources. The very old ones, we probably don't do much with that, but with the new ones, those that were allocated last year, we actually had some work with that.

So I'll show you the beast, the ugly underbelly. Well, that is a formula (refers to slide). We throw in a scoring unit over time, again, time is discounted when the resources have been allocated five years ago or ten years ago. We look at IPv4 addresses, IPv6 addresses, ASNs and so forth and the scoring unit, as I say, a /21 IPv4 size is about the same rate as a /32 in IPv6 and these are all knobs we can twiddle, and this is something that we have presented to our members and they, every year, twiddle it a little bit more, do something here, do something there, but this is basically the formula. And I don't know whether anybody really understands it any more but it works.

So then we want to clear our minds from that and go to the beach and think about clouds for a little while.

What comes out of it anyway are those fees. Again you see the development over time, fees for this year are between 1,300 up to 5,500. We also have an assignment fee, an administration fee for mergers and the like. I should have put it on the slide, I haven't, the number of members we are currently catering for, that's about 4,700 and a bit and by the end of the year we think we'll have just over 5,000 members if everything goes well and according to plan, which it might not. You'll see in the first column in one of Randy's slides, we do run a significant surplus every year. We try not to do this. However, our board has said we also want to maintain a one-yearly reserve. We have more than that and actually what we do this year, or what we did last year is we give some money back to the members.

We didn't give money back, we gave a rebate to existing members, saying, "We want a one-year turnover we currently have 1.5 years or two years. We don't want that much money." We could have dramatically lowered the fees but we want to have some stability. That's why.

And that is, that is the magic. That works for all the time that the RIPE NCC is a membership association and does not rely on voluntary contributions any more. That is at least since 1997 I think.

If you have any questions, I'm happy to try to answer. At some time, I did understand this formula.

RANDY BUSH: Thank you, Axel.

To me there are two very interesting differences here.

One is that your time model - in other words all the resources are discounted and it's a straight-line discount by the way. OK, so something that is four years old counts for half as much as something that is two years old. OK? It's much more oriented towards charging for service. Whereas the current APNIC model very much looks like you're renting IP space. Right? And I think this is very interesting for me.

The second one, if you mind me going through your foils for you, is that the difference between what a small member pays and an extra large member pays is significantly smaller than in APNIC. And that's just another significant difference. I don't know whether that's good or bad, it's just that stands out to me. OK?

HYUN-JOON KWON: I'm H.J. Kwon. I'm from KRNIC. I have some questions about the business plan. Is this still any special committees or any variations of the RIPE today when you set up the business plan?

AXEL PAWLIK: No. As it goes, we write the thing in the Secretariat at the RIPE NCC and we publish it in a draft form and we hope to get support for that and usually we do get support for that, from the community at the general meeting and that's how it works.

We have, three years ago, set up a RIPE NCC Services Working Group within RIPE that discusses the quality of service, the amount of services, the various services that we are providing, should be providing or maybe should not be providing. That working group, as it is an open Working Group, anybody can come really, does not speak about the budget factor, does not speak about money. It's just about quality of service.

HYUN-JOON KWON: So you mean it is discussed in the Member Meeting, right?

RANDY BUSH: I go to both parties so maybe I can clarify. The word 'business plan' is the wrong word, OK? It's the activities plan, OK? And that's very different. When Axel says they don't discuss the budget, it is, "Should RIPE be holding training workshops in the Middle East, North Africa and Iceland?" "Should RIPE be developing software tools for measuring Internet infrastructure?", etc, so it's very much what kind of activities, not micromanaging RIPE's business plan. Very different.

HYUN-JOON KWON: I'm sorry. I made that word, I don't know.

AXEL PAWLIK: No, the word is the 'activity' plan.

HYUN-JOON KWON: So you don't have discussions about the business plan?

AXEL PAWLIK: We could have discussion about it. That's why we publish the documents. There is relatively little discussion about the budget. There is a little discussion about the activities. There is more discussion about the charging algorithm. Should we be charging for IPv6? Should we be charging for IPv6 address allocation services? But the main discussion about services, what we are doing, what we are not doing about activities, are in the open forum in the RIPE Working Group on RIPE NCC services. The budget, more or less, is only discussed among RIPE NCC staff and the executive board, saying, "We heard we should be doing root server operations. Now, this costs us so much. Is that OK with you, the board?" and the board says, "No, that's too much. Cut by 20%," and we cut by 20% and that flows into the draft budget that we publish.

MING-CHENG LIANG: I think this model is to me very interesting. And you mentioned that activities are approved, but I think each activity, the cost of the activity itself would be more or less, since you have been running for some time, so for each activity how much it costs will be reasonably known to some extent, right?

AXEL PAWLIK: We don't - and you can look it up on our website - we don't present a fine-grained budget per activity. We bundle our activities into membership services, that is for instance, mostly registration services. Then there would be coordination services, that is support for the RIPE database, for the software, root server would fall into that. And we do something that we call information services that is communications, WSIS and the like with all the Internet.

MING-CHENG LIANG: And the second question. Is this to mean that - if I understand it correctly - you said that you categorise your members into four different groups, right, five different groups? In these five groups, you decide 20% for the small and then with the relative calculation, with whatever it is, you calculate who should belong to which group?

AXEL PAWLIK: Yes. We start counting at the bottom and we say 20% need to be small members.

MING-CHENG LIANG: So it is like you calculate your list according to maybe your budget and say, "I need something like this."

AXEL PAWLIK: Yes, we have a budget we agree with the board that that is how the budget should be. It's Ä10 million to Ä1 million for the last four years, so it is fairly stable. This is how much money we would like to have. This is how we distribute the cost to the members and then the members agree in the end, hopefully.

RANDY BUSH: Can I amplify that question? Excuse me, Paul. I'm being picky as chair. You do these membership categories by allocation, are those by address holding? And is that... so. What's the metric of size?

AXEL PAWLIK: The size is developed through this algorithm. We throw all the address holdings.

RANDY BUSH: So it's aged address holding. So if I have an enormous amount of address space that I got in 1992, I'm a small member.

AXEL PAWLIK: By now, that's entirely possible. Unless you have received new addresses.

RANDY BUSH: Right. Just checking.

PAUL WILSON: Axel, how are the actual fees, the fee levels, determined? The US$5,500 for a large member, etc, how is that done?

AXEL PAWLIK: Say that we need, let's say, Ä10 million, we have 5,000 members, roughly. We want 20% of them to be small, 1% of them to be extra small. Then we just make it up. We make up some numbers that fit that are not too wide apart, as you've seen, from the fees, that we present to our members at the membership meeting, at the general meeting. And then they note it off.

PAUL WILSON: There is no formula that says?

AXEL PAWLIK: No. We play around, see if it makes sense.

PAUL WILSON: I'm interested in the difference between the lowest and highest fee level. One reason is that, as we've heard, RIPE NCC spends a lot of its money on community services, the K-Root server, you're active in the WSIS, which is all about education and securing, somehow securing, the Internet on behalf of all of the community. But the value of that service, I would think, would be quite proportional to the size of the organisation. If you have a very small ISP, then the value of your services to that ISP can't be bigger than the budget of the ISP, for instance. If you've got a $1 million ISP, you can't say that the value of its access to K-Root would be more than $1 million. But if you've got a $1 billion ISP, then its ability to access K-Root locally, its ability, its knowledge, that you, RIPE NCC, are protecting its interests in the United Nations, to stop some disaster happening which could wipe $0.5 billion off its value. That's obviously a much bigger thing. I'm interested in how the extra small members feel about paying such a disproportionate amount of the costs of RIPE NCC when the value, as I say, the value that may be a $1 billion telco would get from your services must surely be astronomically larger than Ä5,500 a year.

AXEL PAWLIK: We did not see much discussion on that. Initially I thought the big members should be paying quite a bit more. This is how it turned out in the end. There are so many of them and, in the end, if you look at it, Ä1,300 a year is not much money for a small ISP.

PAUL WILSON: Unless you're in India, but sure.

AXEL PAWLIK: Looking at our service region, I think that is what we see and we don't get much discussion on that.

PAUL WILSON: As you said before, you have 5,000 members now.

AXEL PAWLIK: That's projected by the end of the year. Now it's 4,700 roughly.

PAUL WILSON: APNIC has 1,300. I'm very envious, Axel.

RANDY BUSH: So size matters?

KUO-WEI WU: I have a comment about the historical why we set up this - it's not changing every year, because a lot of members, it's very difficult for them to change the payment every year because they need to allocate at least one year in advance. So that is the reason, I think, I believe, that Che-Hoo's historical presentation, that is the reason why we set up the fixed membership fee, but I think that is the reason for the dramatics. Thanks.

AXEL PAWLIK: That's right. That's something we have heard. Unfortunately it has just fallen off the slide, we did make quite a significant loss in 2002. The bubble had burst. And -

KUO-WEI WU: Do your members worry about changing the fee every year?

AXEL PAWLIK: Again, with those numbers, they don't really worry about it, but our board decided to propose a rise in fees by 50%, on average, to replenish the reserves that we lost at the time and we did get some strong comments from some of our members saying, "How dare you raise the fees by 50%!" but looking at the numbers it might have been Ä200 per year again for large member and since then we also have come down in yearly costs, so that is something they like to see - a little bit less every year. And with the astonishing growth in pure member numbers, that is possible.

RANDY BUSH: I do live in Hawaii, as you can see by also wearing the flower, but I come from Seattle, and I care about coffee, so Geoff will be the last.

GEOFF HUSTON: One point of clarification. For APNIC members out there in the APNIC region, because they're paying their fee in US dollars and the US dollar exchange rate changes all the time, in point of actual fact, Kuo-Wei Wu, APNIC fees vary every year. The US dollar amount is the same but, in your local currency, APNIC fees vary every year.

RANDY BUSH: All of Europe is not on the euro.

Back to one point Paul made about the flatter fee structure because of membership size, possibly that is not perceived as so much of a problem in that the economies are not as vertically distributed as they are in this region. Maybe.


Session Two

RANDY BUSH: The closing APRICOT event tonight is 7 p.m. at...? That is what I was given. You thought you were going to be given food, but it is not, it is a search, we go out tonight and we look for the event! The next thing is there is a MyAPNIC and policy flash demo all day at the APNIC services lounge helpdesk. Please visit and find out how you can win an iPod Shuffle. Speaking of winning an iPod Shuffle, if you complete the membership online survey, you will have a chance to be drawn, a drawing chance to win an iPod Shuffle. Also, please show the onsite notice board on the screen and to point out to check it regularly for last-minute changes and updates. I will do that, but I don't know how to find it.

NURANI NIMPUNO: I'm Nurani Nimpuno from APNIC. If you go to apnic.net/meeting/23, it is there.

RANDY BUSH: Do you know where the event is tonight?

PHILIP SMITH: Okay, the reception is at the Grand Hyatt in Pasa Senegal, one of the areas around the Grand Hyatt. How you get there? I presume they are providing buses, you could walk, it is a 10- or 15-minute walk.

RANDY BUSH: And now Maemura-san wanted to say something.

AKINORI MAEMURA: Thank you very much. This is on behalf of the Executive Council. Thank you very much, Randy, for coordinating the working group and session. It is very eventful for the first session. We are now going to have the discussion for the fee structure. The Executive Council has the meeting on Tuesday and we had a discussion with the Secretariat about this issue and it was minded that the fee structure is determined through the discussion among the membership. That is you. Not, it is decided by Secretariat.

It is the Executive Council are reminded to represent the whole membership to make the right input and help develop the plan, help the Secretariat to develop the plan. Any kind of input for this fee structure session is really, really appreciated and indispensable to the very good fee structure. APNIC EC will do a job for the, you know, represent the membership and have to lead some technical group for discussion. But true, the discussion is very precious for that, so thank you very much for your cooperation in participating in the discussion.

RANDY BUSH: Now, I'm going to do something sneaky which is try to sneak in three minutes from someone who does not even know that I'm going to ask them to talk. But I know there are spies here from other regions. Surprise, surprise, we don't know it, usually. But there is another region that has NIRs. That is LACNIC.

Ricardo Patara, are you here? Ricardo, could you speak for two minutes for free, because it is not fair to all those who want to talk but tell us what LACNIC financial and relationship structure is regarding the two, I believe you have two NIRs in the region?

RICARDO PATARA: Yes.

RANDY BUSH: And they are national, Brazil and Mexico? I'm sorry to surprise you. He didn't know I was going to do this.

RICARDO PATARA: Yes, we have two NIRs, it was something new which was implemented in 2003 and 2004. There is an agreement between LACNIC and the NIRs, and our members of the NIRs are automatically members of LACNIC so they can vote, they are part of the board of LACNIC and the fee structure is something like every year NIR send us the number of the members they have and the total amount of revenue on the service provided for the members. 30% of the revenue is sent to LACNIC. This is already implemented with the Mexico NIR, it is going to be implemented this year with the Brazil NIR too. So the most important part is these members are also members of LACNIC. So they can vote, they are participants at the meetings.

RANDY BUSH: It is a flat deal with the NIR? So much percentage of their income?

RICARDO PATARA: Yes, if there is, I was looking for the exact numbers, I can provide it in the mailing list later, I couldn't find it, sorry, but 30% is, actually the amount of the revenue they collect for the service is bigger than, a little bit more, not only the flat rate, 30% rate. I will look for the number, precisely and send it through the mailing list.

RANDY BUSH: And do you have any idea how the NIRs decide what to charge?

RICARDO PATARA: They basically implement, they use the same fee structure we have. We have, like, let me check here, seven categories. They use the same fee scale, structure we have. We have like small, micro, medium, large, extra large and bigger. We call this major. They use the same fee structures to their members. So they usually charge the same fee for the service we have. Basically, our fees structure is based on the amount of address space the members have so they fit into one of the categories.

RANDY BUSH: It is flat? How much time delay?

RICARDO PATARA: No, it is flat. A small member would pay $1,800 per year in the category of medium.

RANDY BUSH: Do you use an external currency as your standard?

RICARDO PATARA: Yes, we use US dollars but incur expenses in other currencies.

RANDY BUSH: You have the same problem.

RICARDO PATARA: Yep, something like that.

RANDY BUSH: It is nice to have perspective from how other people are using it, especially since they have NIRs. I believe I, this slide is incorrect and the next presentation on the NIR meeting report will actually not be given by Izumi but given by Maemura-San.

Is Izumi aware of that?

She is now! We had a surprise speaker a minute ago, now we have another one.

IZUMI OKUTANI: Sorry about the confusion, I thought Billy would be presenting the same presentation that was done at the NIR SIG, so I really appreciate that you have actually secured time for me but it would be pretty repetitive. So I can do a brief update but I think it overlaps. Do you prefer me to speak?

RANDY BUSH: Talk to the microphone, do the NIRs want to give a report for 10 minutes by whoever gives them?

BILLY CHEON: My presentation is the only thing. I will.

IZUMI OKUTANI: I was thinking members will be presenting something, sorry!

RANDY BUSH: No wonder the NIR structure is so expensive!

AKINORI MAEMURA: I understand what she is talking about, but NIR meeting has only one agenda regarding fee, so Billy's one, we will present so, please get rid of the (A). Thank you very much.

RANDY BUSH: We have taken the 10 minutes off you. I'll push the button, ready?


MING-CHENG LIANG: Thanks, Randy. I think that after I heard a special RIPE model, I want to change my models because I think Randy has said we were caught in the address model verses services type of argument. But in this model I presented, it is the base of the rental base as Randy said. What I did here, is that we tried to see, I think from my understanding of APNIC concerns, they have two goals. One is to increase the stability of their revenue so that it is easier to do. And then to have a more, make it fair, more fair to the members.

Of course, from our perspective, the third parties who care, but our perspective is we hope the change will be as minimal as possible. In the previous model, I think we started using what we exactly have right now. The three bits and then you still have to treat the NIR as a whole, all that type of stuff.

So in this assumption, based on something like that, but we make some changes to introduce. Because I think in Paul's proposal, this continuous model it seems to us it makes sense. Because before, I pay much, pay maybe the same as 12, something like that. It might not be quite fair. So we think we can adopt APNIC's model. The result is it seems to be, another thing is that to increase the stability right now we charge it on address-based contribution from NIR and maybe if you were going to increase the stability we increase members from 2003. The reason we chose 2003 is because it is the time, I think from my information, that is the time that you changed this allocation schemes.

Maybe that is not quite right but it is just an idea of timing. And then using an average through the years, then maybe say how much to pay, it is another possibility.

So this, it will be stable. And the difference we can see is that this is the current model, the blue one. And the yellow one is actually what we propose.

The difference is in our proposal what we do is actually just use the continuous model so if you are right on this spot, it doesn't change you at all. But if you are not right on that spot, the /10, it will increase by 58%. If you are /11, it will increase maybe 21%, something like that. But it will continue.

Since it is a continuous model, the upper limit of this is extra large, so it increased after that. So still, some will be impacted more seriously than others.

However, this is APNIC's model. You can see the curve change is quite extreme, compared with this change.

Then we calculate, according to this model, this is what impacted if you hold an address in each mark. It is continuous, so you have to see what it is. But if you are right on the mark, it gives you an idea of what might happen. This one, 125, 26%, the other will be 58% and some will not increase at all because they are right on the spot.

This is data I got from the APNIC version 2.3, but I took an average because I don't know how they get, I took an average. The assumption is this one, 6.62, so we make a calculation so it is an educated guess. The difference it is for LIR. If currently, they are paying about 40, $4,600,000, something like that, and then in this new model they will paying about 5.2 million, something like that. That will be gradually increased because of the continuous model.

Then for NIRs, I think that it is also increased. From this NIRs proposal, for the NIRs membership, this is a list to compare the LIRs and NIRs. NIRs increased about more, about 50% in this membership fee. Then the actual consideration, the actual means, it is still similar to the trace but it is average. It will pay a, it will increase. The total from this scheme, we can have about 15% increase in revenue. Just estimation.

LDC we have discussions with others and think maybe others in the APNIC and NIR can work together to see how are we going to deal with this. I look activity based from RIPE NCC because it will help. Because the activity base, then we can see what we want and then maybe help and maybe NIR can contribute something more. I think this is a possibility.

If, I think that I kind of like, in the RIPE, I like the idea of the activity approval. The activity approval and you need a certain budget. Maybe we can mark it by a factor to meet the requirement, if that is the case. So that is it. Thank you.

RANDY BUSH: Are we ready to have, first of all, any questions? Are we ready to have another go at?

SPEAKER FROM THE FLOOR: I have a comment about, it was a very good presentation and I have a comment about LDC, can you show the page? Yes, actually, Paul's proposal he wants to give discounts to the countries, from my personal experience I visited many of these countries and when I look at the ISPs in the LDCs, most of them are the big ISPs. So helping them and also narrowing the divide is two things. The address ISPs a good way, but a better way is to help them in other ways to have a support team, all the countries, to attend the APNIC meetings. I think it is better to giving, ISPs in all the countries are sometimes much bigger than the medium-sized ISPs so maybe we can find out and give a better way to help the countries than giving discount to LDCs.
That is my comment. Thank you.


BILLY CHEON: My name is Billy CHEON. Before my presentation, I believe NIR and APNIC have cooperated very well and have enjoyed successful years and hopefully this time we can find some solutions for both APNIC's community.

In my presentation, I will like to confirm a couple of things that is to do with NIR. It is rationale from Paul's presentation, I picked up a couple of rationale from Paul's presentation and I will show you some analysis and then I will finish with some suggestions to be included in the future.

Overall revenue stability, it is already done by Paul and I will make it quick. There has been exchange valuations since '96 and for the four years, APNIC activities and services have increased with no operational surplus. I recognise this and agree as APNIC.

Next slide, please.

Annual fee calculation. Instead of according to six tiers, the proposal to introduce a continuous formula, and I think it is reasonable and agree with this.

That is right, please. NIRs. I think this is really a controversial part. I will tell you more in later part of my presentation.

Non-member and historical resource holdings. I don't know about other NIR situations but we have done a project in cooperation with APNIC and it was very successful and I think it is important to do this in consistency of global Internet manage management.

I support this.

I show you the analysis of KRNIC. The figures here in 2005, we paid 229K which consists of 14K of membership fee, 180K of address fees so 20K we paid in the year 2005. So I also include the figure in the 2006, but 2006 we paid around 200K so I think average can be 200Ks in US dollars. The figures in the box, this is payment that we should pay once we adopt version 2.3.

As you see, KRNIC have to bear NIR fees from 40Ks to 227K and also we, instead of paying the per-address fee, we have to pay 285K for NIR members' fees. So 413K, it is a double increase from payment. So it is really difficult for us to bear. I'm wondering if there is any organisation who can survive if there is 100% increase in amount.

So we projected what it would be like in the future. As I mentioned, the first year, we adopted version 2.3 it goes better and then at the end of the year we have to pay more and more because of we get allocation from APNIC.

Next please. Now, I will tell you what our conclusions and suggestions. As I mentioned, I pointed out during the first presentation if you see this picture, I think it seems very unfair. See, NIR and NIR members has almost 50% of IP address right, then they are just paying less than 20%. It seems really unfair. However, as I said, please notice there is another cost behind this here somewhere here. All NIRs are already paying. We need to build up database and need to hire host masters for local matters. So those things I want people to recognise this fact.

In general, I, KRNIC agree with version 2.3. Also I believe and I should expand contribution in APNIC region considering responsibility of NIRs for AP community. However, there is a couple of points that I can - we cannot accept.

This, according to version 2.3, charging scheme is, charging mechanism is separate for NIR and NIR members. But I think we cannot accept it because NIR members is not APNIC's member. Right? The proposal separates based on direct allocation but direct allocation is not membership agreement. So NIR members is changing the contract with NIR, not APNIC. So from my understanding, APNIC's membership fee cannot be applied on NIR's members.

Also, according to APNIC, one of three operational policies for NIR in the APNIC region, they ask for transfer of membership between APNIC and NIR. According to this article, they ask freely to be member of NIR or APNIC. It means NIR's members and APNIC's members are totally different entities.

But if the membership fee is applied on a member of NIR, then our billing is like, NIR is not nothing but bill collector. That is what we concluded.

Another point is NIR fees increase too much. As I said, what we would like to suggest is NIR and its members should be treated as one body. That is our voting right. And another fee should be increased gradually. As I said, I think it is really difficult for us to bear one, double increase in one year. It is almost impossible.

Activity, possible activity plan might approval beforehand. So this is very primitive idea at this stage but I would like to suggest first setting up the total APNIC budget based on activity plan and set up an NIR portion out of the total APNIC budget and each NIR pays out of NIR portion based on a number of their own IP holdings.

Actually, there is, during the NIR SIG, there was another point about voting, but our chair commented that this is not the place for the voting issues so I leave it out. This is all for my presentation, thank you.

RANDY BUSH: Somebody have comments on that?


OK, so now we have an open discussion. I would suggest in comments we try to maintain a sense of humour. We are all here to help our customers move packets. I like the way Paul looks at this with fairness issues.

Izumi-san, are you heading for a microphone?

IZUMI OKUTANI: No.

RANDY BUSH: No! Coward.

MING-CHENG LIANG: Hi, Randy, and everyone.

I think that, from today's presentations, that, of course, that, in this new proposal or other proposals, I think there are other things that we can discuss but I think maybe the first thing that we might need to think is that, in our current APNIC model, we are talking about a rental-based address problem. The RIPE NCC model - actually, I kind of like the RIPE NCC model - it's a service-based model and I think it's something maybe we should consider, whether the members want to - whether APNIC will maybe think about a proposal based on this type of fee structure or not. It could give us more chance to have a good discussion from this.

KENNY HUANG: OK, Kenny Huang. I'm not a major in mathematics but it's happy to know all this kind of information, because we've been looking at all sorts of variables in the models. So I'm not going to discuss about the model. I'm going to discuss about a decision model. What sort of decision model will be made? For example, we have been given so many fee models and what sort of decision process will be made by, like, RIPE, made by EC? Or made by APNIC? Or be made by member voting? I'm interested to know the decision model so we can understand why we are here and just sharing the information, so we have something can be doing further. Thank you.

RANDY BUSH: Um, I'm making the decision tonight at 10 o'clock in the bar. I think Paul has to speak to that but I think, like most organisations, we have an EC, is the answer. The board of directors is the EC and they are responsible. But I think the EC in this organisation, like many, is known for having ears, and I think the purpose of this is to fill their ears.

CHE-HOO CHENG: I think I need to clarify. I think the model will probably go through the voting process from the members. Of course for the proposal, which to be put up for membership voting, will be approved by the EC first.

And I want to stress one thing. I don't want you, you know, I don't want people to misunderstand. The proposal that Paul put up is not his proposal, is not APNIC Secretariat's proposal. It's a collection of ideas of many others, including the EC members, including me. And so I don't want people to have a misunderstanding on this part.

Anyway, back to the decision model. It will be put up for membership voting before putting into effect.

HYUN-JOON KWON: My name is Hyun-joon Kwon from KRNIC. I was trying to ask during Paul's presentation and then I couldn't do it at the time. Right now, I have questions to Paul's proposal.

In the slide, there is a comparison, you know, the bar with IP holdings and, you know, we have 15%, NIRs have 15% and then ISPs have 18% of contributions made. Are ERX included in that or not?

PAUL WILSON: No.

HYUN-JOON KWON: It's not. OK. The other thing also is I want to know - without any income - I don't know the exact date - without any income to APNIC, how long can you manage to operate APNIC?

PAUL WILSON: One year.

HYUN-JOON KWON: One year. Right.

Also - I have many questions.

Also, APNIC have a problem with the exchange rate, so I want to know how many percentages of the whole APNIC budget. Also, I presume there will be a natural increase of income, because IP holdings will be increased, so also I want to know how many percentages of the APNIC income because of IP holding increase and then maybe we can find some offsets, or maybe. I don't know how much. If you can answer to that.

PAUL WILSON: The answers are a little longer so I'll use the microphone.

On the exchange rate, the chart that I showed before showed, I think, that the Australian dollar/US dollar exchange rate had shifted from - had shifted by 50% over -

HYUN-JOON KWON: 15? One-five?

RANDY BUSH: Five-zero.

PAUL WILSON: 50%, yeah.

Have you got it there, Randy?

RANDY BUSH: Moi?

PAUL WILSON: You'll need to hit the button there.

RANDY BUSH: This will take a minute. It will undoubtedly cause my laptop to reset.

PAUL WILSON: OK, so the exchange rate on the right-hand side was down 50 cents and it's up at very close to 80 cents now. So the difference there is actually greater than a 50% difference between the exchange rate at the lowest point in 2001 and at the highest point in - well, in fact, 2007 is likely to be the higher one. So, in fact, that makes a huge difference in terms of the translation from US dollars into Australian dollars, between 2001 and 2007, 2006, 2007.

The second question - and the graph is good here - the second question is about the future of APNIC budgets and revenues and it's quite correct that APNIC keeps growing and so the revenues will keep growing, but I think the chart shows fairly well that, in 2003, 2004, 2005 - and you will see tomorrow, the 2006 results and the budget for 2007, that we've got both revenues and expenses increasing and we have, at the moment, we're hovering around a balanced budget with zero surplus, so we have a few years of experience there and we can project forward and possibly we can assume that revenues and expenses will keep going along on the same lines projected forward and still leave us with a balanced budget in terms of the current projections.

RANDY BUSH: Can I insert one extraneous comment? I was on the founding, the starting board of directors of ARIN. And we set as a target a two-year cash in the bank. I don't know if that still holds, if somebody from ARIN is... here. Leslie or somebody speak to that? To the best of my knowledge, the board is still at a two-year operating cushion. It's not important, just to say that one year is on the low end of what I'm aware of around the world.

HYUN-JOON KWON: Which current company are you working for?

RANDY BUSH: This is ARIN. The American Registry for Internet Numbers. Right.

HYUN-JOON KWON: OK. So thank you for your answer. That solved my problems.

Also, you saw in the KRNIC presentation, we understand that KRNIC should expand our contribution because we have lots of IP holdings, but when you manage one organisation, you cannot - just one, you cannot increase 100%. We think we need to have a transition period. So that's why his presentation had the gradual increase will be proper for the NIR position.

Also another thing is about - maybe chair - I don't know if you had a discussion about voting rights about fee. I think the voting is not, is also inseparable, because, when you set up a policy there are some intentions. The fee schedule, if you look at the fee schedule, there is a fee schedule for the members' fee and the voting rights are set up according to the membership. If we are talking about this schedule right now, also we are thinking about changing members' tiers and then that is directly connected to the, you know, voting rights together. So maybe this is a Working Group so this is OK but, when you, you know, change the policy, you are also thinking about changing the voting rights together and that's my comment, thank you.

RANDY BUSH: I - I'm sorry. I interject one comment - one thing I liked about Paul's presentation on the straw poll of, you know, people saying how much, is one interpretation of it was, aside from giving some money to the less developed countries, every one of those votes was, "I should pay less, he should pay more." So I have some sympathy the NIRs don't want an increase. I have sympathies that big members don't want an increase. I have sympathy that small members don't want an increase. There is no free lunch.

KUSUMBA SRIDHAR: I'm Kusumba Sridhar from India. This question is for you, probably. In 1996, we had a review. In 1999 we had a review and in 2006 we're talking about a review and in 2011 we should be talking about another review. One quick question, I was trying to put some metrics together. What's an approximately percentage of cost that goes towards APNIC training and the staff cost in the budget. Any idea on that?

PAUL WILSON: The staff costs are around 45%. The training - I don't have that figure at hand, I'm afraid. There will be some - during the financial reporting tomorrow, I think, in the APNIC member meeting, I think that will be explained.

KUSUMBA SRIDHAR: Extra large - 7,260 in RIPE, $1,716 in RIPE for the extra small. 80% of their costs goes towards their staff and direct expenses. 93 staff versus 1,300 members of approximately APNIC members, 50 staff at APNIC. I think that clarifies the current increase in the membership fee that we're talking about. There is definitely a load bearing on the NIRs at this moment in time. As you know, we have expressed an interest to be also an NIR in India, so we are trying to draw members close. Some members will be published from the APNIC side in terms of training and staffing expenses would be helpful in the future.

PAUL WILSON: I'm not sure I understood every aspect of your point but the financial reporting each year at the member meeting has followed a fairly consistent format over the years. I'm not sure that it will - I'm not entirely sure that it will answer all the questions that you're interested in. Certainly that reporting can be adjusted if the membership is after more or less or different details. I hope that helps.

PAUL WILSON: I had one other point that I wanted to make in response to Billy. I think in selecting a particular fee structure, it's important for us to decide a fee structure based on rational and sound measures. But then to address the implementation of that fee structure is a separate question and I think it's understandable that, if, in a change to a new structure, there are dramatic increases for any member, thank then, in the implementation of that fee structure, you might decide and this has been discussed in previous meetings, you might decide to introduce that change in a phased manner over one year, two years or more. But I think we could and we should - any change to the fee structure should be from the implementation, exactly taking into account the problem that you mentioned.

It's difficult to understand in your presentation that when you say there should be gradual increases - it's difficult to understand that there are gradual increases towards a goal with no system in mind. I would be interested to see what specific proposal you might have that would introduce the change that you desire and that, of course, there could be a gradual, phased adoption of that approach if that is decided.

CHE-HOO CHENG: I have a question for Billy. Do you think the proposal from Dr Laing with the gradual approach, using the three-bit, two-bit for that? Is that... is that the approach that KRNIC can accept. Because it's more gradual than the two-bit approach.

BILLY CHEON: No. I haven't thought about Laing's proposal. I've just seen it. I need more time to analyse it.

But, I don't think I can give...I can say his model is, you know, close to my gradual increasing model, yeah.

PHILIP SMITH: Philip Smith from Cisco. Um, I suppose I'm more an interested bystander in a lot of this discussion, because I'm not part of an LIR or NIR member. So it's just more sitting on the edge trying to understand the whole discussion and all the different proposals, the different variations and the different themes of the different proposals. What to me looks like ever-increasing, complex type solution to what to me seems to be a very, very straightforward problem. Maybe I've spent far too many years sitting in front of routers trying to make the Internet work. Our maxim there has always been "keep it simple". And while we seem to be heading in a very, very opposing direction to the "keep it simple" direction that certainly a lot of Internet engineers have been advocating.

I have huge sympathy for more or less random budget that the NIRs are faced with every year. When you're trying to run an organisation, actually, you like to know what money you're going to have to spend and being faced with this, well, number that can depend entirely on your membership needs, I don't know, I really don't know how you can try and handle that.

A few weeks ago, I did actually propose on the Working Group mailing list that we revert to something nice and simple. There seems to be a cost attached with, I suppose, distribution of Internet resources for want of a better word. So there is a cost, you know, it's a service cost to APNIC, whatever. And I would like to suggest, just in the interests of keeping it simple, that everybody throughout the entire region, whether they're an NIR member or an APNIC member, pays that same cost. Now whether the cost changes for the size of address space, I have no real comment on that, but I would hope that we could consider that, you know, if an NIR member needs a /19, they pay the same cost as an APNIC member who needs a /19. Now, I can see all the extra vibe that the NIRs bring to their membership, the local language capabilities which are a big advantage.
They're closer to the Internet community, the actual Internet communities, the Internet resources. So the terminology I used in my posting is they provide a value-add over what APNIC can do.

In typical, in many other parts of the industry, people who provide value-add charge for that value-add service. So I really cannot understand what the difficulty is with APNIC saying, "Well, look, if you need a /19, this is the service fee." And if NIRs provide an extra value-add then they charge extra for that. I mean, there have been responses saying, they can't do it, it's extra cost and burden, but you know, if you go and buy a product directly from a mass-market person, you pay one fee. If you buy it from a value-add re-saler, you pay more. It's how you want to do it.

I live in Australia. If I want to buy a product in Australia, I can pay one price in Australia or, if I happen to be visiting the US, I can probably buy it cheaper there. But I have to get there and I have the additional inconvenience of the longer distance. I see the relationship between APNIC, APNIC members, APNIC and the NIRs and their members in a similar sort of fashion.

So moving forward I can see that as a massive simplification to this whole complex process. Formulas have been put on the screen and discussed and their graphs and sharing this, sharing that and others. There is a lot of mathematics and statistics which just makes my eyes glaze over. Can we maybe consider trying to rewind a lot of this and look forward to some kind of simple solution?

The second thing is just a quick comment about the voting bit. I suppose I can see how the voting would be related to some of this but, again, if you have the end-user of the Internet resource having a vote on how it is actually done, then, again, that simplifies it rather than some block vote type thing the NIRs have, the actual membership get involved directly in the APNIC process. They get involved and vote on it directly. They can see how that could be an option there.

There was something else I was going to comment about but I've forgotten what it is. I will leave the mic to somebody else.

RANDY BUSH: Philip, can I ask you a question? A series of questions?

What do you think about the time-ageing scheme a la RIPE? Does that have appeal?

PHILIP SMITH: It looks complicated.

RANDY BUSH: The formula looks complicated but the idea of the longer you've held the resource the less work you are causing for the registry?

PHILIP SMITH: Yes, I mean I agree that that seems reasonable to me, yes.

RANDY BUSH: OK. You pushed on the question of the cost to me should be the same whether I get my goods from the NIR or from APNIC, which seems to be the LACNIC model. Am I correct, Ricardo? That you pay the same, that the charging model of an NIR is the same as the charging model of LACNIC?

RICARDO PATARA: Yes, yes.

RANDY BUSH: Is that the kind of thing you're speaking to?

PHILIP SMITH: Yes, yes. It's a simplicity thing.

RANDY BUSH: OK. And in LACNIC, those people have, the people who are members of an NIR are also automatically members of the LACNIC?

RICARDO PATARA: Yes.

RANDY BUSH: So, and do the NIRs have a vote? Or just all members have votes?

RICARDO PATARA: All members have votes.

RANDY BUSH: Is that the kind of thing you're heading towards?

PHILIP SMITH: That is the direction I was heading towards, yes, where it's everybody participates in the actual APNIC process and everybody has an actual say in what the fees are and so forth.

I'm not saying that the NIRs don't tell the membership what is happening. I'm not saying that for one second. But at least this way, with the one-member-one-vote, central participation, everybody feels part of the community.

I've remembered the third part. Can I address that? And that was about the developing countries versus the developed countries. Some of you know I travel an awful lot and one of the biggest burdens I see in the less developed regions of the Internet is actually meeting the APNIC fees. US$1,000 for Australia, Japan, Singapore, whatever, is lose change. US$1,000 for Bangladesh, Nepal, can represent a month's income for a service provider, a year's income. It's a massive amount of money for them. Therefore it was Paul who suggested something based on the UN definition of LDCs. I would really, really support that in any revision of the fee structure. I see it as the biggest barrier to entry for a large part of the region and, if we can get away from some of the difficulties that these ISPs have, most of them can get two addresses and I mean literally two addresses and the rest of the entire address space behind network address translation devices double NATs and triple NATS and all other mess. It doesn't make the Internet work for these people.

RANDY BUSH: I have great sympathy for that. I work at the low end more in Africa than here but the entire Uganda for years ran on a /27. One entire country and it was 12 ISPs at that point. In some countries, the US$1,000 fee is the cost of two engineers. So I have real sympathy for that particular point.

Maemura-san.

AKINORI MAEMURA: Can I show something?

RANDY BUSH: Please. Or that remains to be seen - whether you can or not. Ready? Push the button. Well, we have the answer. Ha! There you go.

AKINORI MAEMURA: Akinori Maemura from JPNIC. Philip has said his thought for the NIR system in the Working Group mailing list and I had some response. I still feel like something is missing in the understanding on the regular member regarding the NIR business.

So this is one of, you know, the illustration to look at what is NIR job. I'd like to point out that the NIR is doing - apart from the registry - they're doing allocation assignments and second opinion requests from the members, the collection of fees, the inquiries from LIRs, the member meetings and the training to the LIRs. And APNIC is doing, of course, the common address pool management, and inter-RIR and ICANN coordination and the inquiries from and to NIRs and the second opinion requests from NIRs.

So this portion is, I suppose, this portion is a more heavier here than regular membership people think. I may be wrong. So please point out if I am, if I am wrong.

So, is this just the same as you supposed, Philip?

PHILIP SMITH: This is great. This is the kind of thing that I was really looking for to try and see. And so can I suggest one thing? Can the NIRs and APNIC work out between themselves what the costs of the relative bits are so that the value, so that the charge to the end-user is exactly the same? Because then the NIRs will be able to work out, OK, so maybe 50% of the service fee is the NIRs and 50% of the service fee is APNIC's. But the end-user, so the person getting the /19, pays the same right across the entire region. Because this way, you don't have to pay this - to me, I'm simplifying but it looks like a random fee to me and I don't like running businesses with random fees.

AKINORI MAEMURA: Alright. Thank you very much.

RANDY BUSH: Philip? Spoken from Cisco?

LAUGHTER

AKINORI MAEMURA: I want to ask other people what is the case with your RIRs. Ricardo?

RICARDO PATARA: Sorry?

AKINORI MAEMURA: Do you find any difference from your cases in the LACNIC region? In this illustration?

RICARDO PATARA: Not much. One important thing is, in our case, the language barrier is not so big as in other regions just at this point.

RANDY BUSH: What's interesting, though, is he has a data point for Philip's question, which is 30%. Somehow they have decided - is that always - is that figure reviewed? Or has it been 30% since 1910 and will be 30% forever?

RICARDO PATARA: There is no planned change to that. No discussion for change.

AKINORI MAEMURA: So, yeah, I heard that 30% is your case and right now, in the case of JPNIC, it's like a 10%. in our case. So I am now looking for what is the difference. Now, this is just only for the NIR case and I have one slide.

In the previous slide, I tried to figure out what is the cost factor in the case of NIRs, but we would like to work on this kind of analysis between big and small members, so, you know, Paul's analysis is basically based on the service and cost. A service versus cost kind of analysis. But, I still feel like having some cost factor to fee kind of analysis. So then the cost balance between the bigger members and the smaller members will be justified. What do you think? That's my comment. Thank you.

HYUN-JOON KWON: Hyun-joon Kwon from KRNIC. I have some comment about the membership thing because LACNIC introduced in the NIR members, also it's treated as, you know, the LACNIC members. In APNIC cases, I want to - I came here to - you know, in our APNIC policies, we have an operational policy of the NIRs and according to the policy, there is two different memberships definitely so there is also the membership between APNIC and the NIRs so we cannot treat them, you know, the NIR members, as APNIC members.

So also, to make any fee schedule for non-members, phoney members. So you know, that's why we're thinking NIR and LIR members should be treated as one unit according to all the policies.

RANDY BUSH: But it's - so let me be sure I understand. You are saying that, because NIRs' members are not members of APNIC, that we should not be saying that the cost to NIR members should be the same for the same kind of thing. Is that correct?

HYUN-JOON KWON: OK. What I'm saying is exactly, we know that according to the new proposal, they are applying to the fee schedule, to the new members' tiers with the continuous formula and then they calculated that what they charge the new proposal charged our individual NIR's, ISPs. But, I'm suggesting that the NIR's member is an IP member so the NIR members' IP holdings and our holdings altogether should be treated as one. So just to treat it as a large NIR. So just to treat it as one member. That's my comment.

CHE-HOO CHENG: I don't know I fully understand what he said. But anyway... I think the model of version 2.3 is still the members of NIRs still are not - the members of NIRs still are not the members of APNIC. So to APNIC, they are still only an NIR. So they don't - the only difference is to calculate the fee that NIR pay to APNIC is according to, you know, the IP address and also the - the IP address of each NIR's members or something like that. Very complicated.

Anyway, talking about the work that Maemura-san just raised regarding the cost distribution between APNIC, and maybe NIRs, and then determine something like that, I want to say that there should be some consistency among the NIRs. I don't think, you know, for each NIR there should be a different discount. Think it's better to have one discount level for all the NIRs. but because your situation in JPNIC may be different for this issue from other NIRs. I don't know how they do it but anyway I think there should be consistency.

OK, I'm just giving you some random thoughts. In fact, I think I mentioned it already, if the model is set up right, there may be a situation that the existing LIRs from, let's say, Japan, will change to JPNIC and then JPNIC can have better economy of scale, better - more income and things like that. Can you look at it, look at the model this way and then it will be an advantage to you. I don't know.

OK. Maybe later.

RANDY BUSH: Chris?

CHRIS LILJENSTOLPE: I'm Chris Liljenstolpe from Alcatel. I've been in the region for a couple of years. I mainly help with the larger carriers. However, what I'm up here to speak about is to reinforce what Randy and other folks have said. Is that I do see the charges for address space here being a major cost, especially to the smaller competitive carriers that may want to spring up in region. The pricing, I think, to a certain extent, reinforces the large carriers' positions where they can take income from other operations and use this to fund address space and other things. So it might be prudent to look at the lowest-developed countries introducing some kind of sliding scale for address space maybe based on cost of - engineering costs, per cent of engineering costs or something along those lines, to enable - to move that one high hurdle they have to cross.
They have other high hurdles they have to cross as far as connectivity etc, but do we really need to make as much money from, say, a small operator in Bangladesh or say in Cambodia than we do from a large operator in Japan. I think maybe introducing a sliding scale for the least-developed countries would be a useful thing for the industry and the region.

RANDY BUSH: OK. Have about 15 minutes left. I care about lunch almost as much as I care about coffee. And tomorrow you will notice that we have one whole hour at the general session for this discussion because it is so important to the membership.

But I will have the task of reporting on this meeting and so I would like to ask a couple of questions to get a feeling of the room that is not a vote. It's just - I'm going to ask how many people are interested in certain approaches, things like Axel, for example, how many people think the RIPE model of the time curve that the longer you've held it, the less it counts, is interesting and we need to think more about it. How many people actually think that? How many people think it is not? How many people don't think? It seems a large number.

How many people think the idea that it should cost the same if you are a member of an NIR versus if you are not is a useful concept? In other words that it costs the same - aside from the lower developing countries' issues - how many people think that all stores should be charging the same price? Do people have a feeling that that's an important issue? Are there some people who think that is a very bad thing? Again, the people who don't think win.

I don't want to get into some of the other issues, because it's pure politics.

Maemura-san.

AKINORI MAEMURA: I'm afraid I don't still understand - I don't yet understand your question. Sorry.

RANDY BUSH: I'm trying to find out what things need to be discussed more. Where we should put effort. Should we put effort into discussing the RIPE concept of applying time to the model, which is not there at all and is in none of the proposals? Should we spend effort on that or should we say, "No, that's a stupid idea for Europeans and we'll continue the way we are going"?

CHE-HOO CHENG: I have one point regarding this. I am - yeah, I think this is interesting to look at, but I think it would be much fairer if the historical address space was also included in the consideration.

RANDY BUSH: One of the things I like about it is under that scheme it makes it a little fairer, maybe, for those to be considered. To me, that's a very difficult issue. The term we have for it in the States is 'legacy address space', the old address space you got before the registry, right? You got it from SRI or whatever. That's a very difficult issue because, in my culture, doing something for somebody who is not here to represent themselves is, you know, we had a revolution supposedly because of that. Right? So it's a tough one. What I like about Axel's time thing is this helps solve that. This helps solve it. It can bring the legacy address space problem and the ERX in a model that maybe is fair. So I'm agreeing with you. I'm agreeing with you strongly but saying, why?

MING-CHENG LIANG: I think that maybe Che-Hoo and APNIC's concern is that some people are holding that address and don't use it. Maybe they are thinking about them. But I think there may be a separate issue that should be dealt with and no-one we might think up some way to reclaim or something like that. But - I think that the service model by RIPE NCC is an interesting one and it should be - personally I think it should be looked at more closely.

RANDY BUSH: Other people on that question? Izumi-san.

IZUMI OKUTANI: I just want to clarify why I didn't, like, raise my hand either for or against. Because I like the idea of time-based because it tries to consider the actual cost base in it but, I'm not sure if we should base it on time or other cost factors. So I like the idea of what, if anything, some kind of fee based on the actual cost, so I support that broad idea.

RANDY BUSH: Service cost as opposed to rental space?

IZUMI OKUTANI: Exactly. I'm not sure if that should be based on time or other factors.

RANDY BUSH: Good point. Good point.

MING-CHENG LIANG: Could I get one more comment because in this NCC model, I think it's relative, not like in our case. Up til now, it was arbitrary. If it's small, it's small. But it's relative. So I think it gives us more room to do something, since it's relative. I would think that this is something to consider, because it's relative. My constant problem is that you might have a problem with the fee change every year. It might be something like that. And that's the only thing I think is bad. But other than that, I would think that maybe there's a time decay or other factor. It's just the parameter in calculating relative compared with this whole society. So it seems to me to make some sense.

KUSUMBA SRIDHAR: I think, in my view, I think LACNIC more or less is quite close because they have NIRs like APNIC. And RIPE don't have any NIRs in their region. Probably we should also look at the LACNIC model more closely than what we look at the RIPE. Because RIPE - the numbers are significantly different than APNIC or LACNIC.

And I want to seek clarification that you're collecting the membership fee quarterly and half-yearly also?

AXEL PAWLIK: That's an option.

KUSUMBA SRIDHAR: Probably that is something that is interesting in terms of membership - collecting the membership quarterly and half-yearly. We all know that, when you put in the mailing list, we know how many bounce back because most of the members, especially - we were just commenting that, in the APNIC region, most of the members are government members who basically buy this almost a year or one-half year in advance and some of them keep it and don't even use it. They only look back at APNIC only when they get the invoice. When they get the reminder for the payment and the second reminder saying, you will be disconnected or your resources, will be pulled back. So, if there is an option of quarterly and half-yearly payment options so the member also looks at things frequently too, it will not be a load on him to pay all at once as a single year, I guess there will be a little increase in the costs of following up and getting the payments on the APNIC side but I think that is worth doing, which will have a double advantage of getting the member looking at APNIC more and also the feeling that he is only paying quarterly or half-yearly which reduces the single load.

That's what we also need to take from RIPE while looking at LACNIC.

RANDY BUSH: Excuse me if I - I do not see the RIPE and LACNIC models as in conflict. I could see - and if people weren't going to talk, I was going to make trouble by suggesting a model that was a combination of the both, that said aged time delay - though I hear Izumi-san too, that we should look more at the service that's provided and that may not be purely time – but, some decay and see that it's charging for service and the idea of flat, the idea of everybody's a member, the idea of fixed per cents. I think there's a combination in there but I don't - it appeals to me because, like Philip, I'm an engineer and I like simple, OK? And one formula that has a sigma in it does not scare me because I can read the math, but it was a simple formula really. But I don't know, you know, there are players here - Che-Hoo and Maemura-san, etc, etc, with much longer history and much more understanding that this may be crazy.
But I wonder if there is something in this space that will make everybody happy and be fair and can go forward? And also provides a stable model for APNIC and the NIRs and the individual members. I don't - so I don't see the ideas that we got from RIPE and LACNIC as in conflict at all.

IRENE CHAN: I am Irene from APNIC. In response to what you said now. APNIC do offer payment options for quarterly and half-yearly, but there are additional cost to cover registration fees. We normally offer to them when they say they have difficulty in meeting the annual membership fees.

RANDY BUSH: I can't believe this group will finish early, even if it's only five minutes.

Thank you very much. Thank you all for maintaining friendliness and a sense of humour.

(End of session)