Your IP address could not be determined at this time.

ARIN XIV Public Policy Meeting Minutes, Day 2 - 21 October 2004

ASO Memorandum of Understanding Signing Ceremony

Speaker: Ray Plzak, ARIN President & CEO

Ray welcomed those in attendance and introduced those who would be involved in the signing. Signatories included: Raúl Echeberría , LACNIC CEO; Axel Pawlik, RIPE Managing Director; Paul Twomey, ICANN President and CEO; and Ray Plzak, ARIN President & CEO. Ray then invited Vint Cerf, Chairman of the Board of the Internet Corporation for Assigned Names and Numbers (ICANN), to make remarks.

Vint thanked Ray for the opportunity to speak and said that the ICANN Board is deeply appreciative and has high regard for the work of the Regional Internet Registries. He noted that this new MoU will cement a framework for global Internet address policy development and it represents an important milestone in the evolving mechanisms of Internet policy development. Vint added his personal thanks to all of the RIRs and to everyone in the audience, remarking that many have devoted whole careers to making sure the Internet works, and that nothing is more important than ensuring that it continues to operate for the nearly one billion current users and the billions of future users.

The signatories commenced with the signing of seven copies of the document. Ray noted that one would go to ICANN, one to the Number Resource Organization (NRO), one to each of the four active RIRs, and one would be held in trust for AFRINIC. Ray then thanked all those in attendance and said he looked forward to working with ICANN in the future to make their relationship even stronger.

ASO Signing Ceremony

From left to right: Ray Plzak, Paul Twomey, Axel Pawlik, and Raúl Echeberría.

Call to Order and Announcements

Speaker: Ray Plzak, ARIN President & CEO

Presentation (Read-only): PDF PPT

Ray opened the second day of the ARIN XIV Public Policy Meeting at 9:30 AM EDT. He thanked America Online and Time Warner Cable for their sponsorship of the meeting, including Time Warner Cable's sponsorship of the ARIN XIV social the previous night.

Ray offered information on the Terminal Room/Learning Center, the ARIN Help Desk, and the ARIN XIV meeting survey. He provided information on the success of the remote participation effort for this meeting, with 2 participants in the WebEx chat, 22 RealMedia webcast viewers, 21 multicast-MP3 audio users, and 5 multicast viewers.

Ray announced that Sanford George was the winner of the ASO AC election.

John Curran, as Chairman of the Board, moderated discussions throughout the day.

WSIS and WGIG Feedback

Speaker: Ray Plzak, ARIN President & CEO

Ray opened this section of the agenda by saying this would be a chance for a general discussion about the World Summit on the Information Society (WSIS) and the Working Group on Internet Governance (WGIG). Ray provided background information and then opened the microphone for comments from: Paul Twomey, ICANN CEO; Scott Bradner, ARIN Secretary; and Raúl Echeberría, LACNIC CEO.

Paul, Scott, and Raúl delivered personal remarks based upon their experiences in the political arena that WSIS encompasses.

Ray encouraged everyone to participate in the discussion, noting that the purpose of this session was to solicit feedback and impressions on what is happening with these organizations.

Comments:

  • Recently, we saw an announcement that the NRO is going to donate money to the WSIS process and I would like you to elaborate on that. Was this to buy a seat at the table or are we just going to feed the sharks more? Ray Plzak responded that the NRO was not trying to buy a seat at the table. The donation was a response to a request for contributions to the Secretariat of the WGIG, which is not funded. The NRO thinks it is important for this forum to be conducted and for all ideas to be presented. The NRO did not want to see a lack of funds being a hindrance to this effort, and so contributed to the operation of the Secretariat.
  • Because of the importance of this discussion to our ability to discuss and develop policy, everyone should solicit the involvement of your vice-president of government policy or your general counsel and get them to pay attention to this issue. The business community, which has been so important in the growth of the Internet, is not represented directly in the WSIS process. It is currently a government-oriented process, and Internet companies, trade associations, and think tanks need to talk to their national governments to let their feelings be known. This is a fundamental threat to the people and operations that you've come to like and respect, and feel are a part of the same community.
  • I would like to remind everyone that when they vote in the elections to fill positions on the ARIN Board and AC, they should remember that these larger global issues will be impacted by their vote, not just localized ARIN issues.
  • What will happen if the report comes out and says that standardization should be done exclusively by the ITU or something equally silly? Do they really have enough power to change anything in the world by issuing such a report? What power and credibility can they use to enforce such a thing? Scott Bradner responded that the answer was both yes and no. The ITU, per se, does not have the authority, but there are many governments that take ITU recommendations as input into laws. When presenting about the IETF to the ITU, one of the questions I was asked was: how can the IETF call these things standards if no governments mandate their use? There are reasons governments mandate standards, usually to protect their local businesses or for political reasons. A change to more government-centric control of standards wouldn't necessarily change what happens on the Internet, but it would add a lot of duplication of effort, a lot of wasted money and resources, and a great deal of confusion. I hope it doesn't happen, but I fear that it might.
  • Vint Cerf spoke and began his comments by saying this is a serious threat. The issues at stake are not about technology - they're about power, control, and who gets to make decisions. Let's make sure we have solid arguments for showing why the current process is the best way to coordinate the interests of all the players in the Internet game. ICANN isn't perfect, and it can still use plenty of evolution and improvement, but we've tried very hard to let every voice speak and participate in the creation of policy. That is why this MoU signing this morning is so important. I keep trying to get the parties involved in WSIS and WGIG to understand that their focus is too narrow. There are some really tough problems associated with the way the Internet is used, like fraud, abuse, and spam. These are areas in which ICANN doesn't have responsibility and technology may not even be able to solve these problems. These are the kinds of areas that could benefit from international cooperation.
  • This is not just the work of other countries. The U.S. FBI, DEA, and other agencies requested that the FCC require new Internet applications be checked by the FBI before they could be deployed to make sure they could be wire-tapped. While the FCC turned them down, government agencies and regulators will continue to try things which are technology atheistic, either for control or to protect revenue streams.
  • I bristle a bit at the tenor of the conversation I'm hearing. Terms like "threat," "battle," and "war," seem overly adversarial. As the Internet grows, more and more communities, each with its own interests, are included. We need to be able to include them. Otherwise the Internet will fail and become marginalized.
  • There have been a lot of comments so far from different parties who have different points of view on this, but all seem to agree that this is an issue we need to follow and participate in. For those who may not know how to keep informed or become involved, it is important for those directly involved in WSIS and the NRO to communicate to the community and the media so that interest can be mobilized. Ray Plzak responded that ARIN will be spending far more funds to promote announcements, newsletters, and press releases in the ARIN region. The NRO Communication Coordination Group will be taking on a larger role in this regard. However, in the end, it is up to the efforts of individuals to take the initiative and read what's being published and find the appropriate channel in which to provide input.

Policy Proposal 2004-5: Address Space for Multiple Discrete Networks

Introduction: Einar Bohlin, ARIN Policy Analyst

Presentation (Read-only): PDF PPT
  • Template submitted - August 19, 2004
  • Posted to PPML - September 9, 2004
  • Staff impact analysis - October 4, 2004
  • Legal review - October 7, 2004
  • PPML Summary:
    • No formal discussion on PPML

Presenter: Bill Darte, ARIN AC

Bill, on behalf of the ARIN AC, introduced the policy, explaining that this streamlines current policy and introduces a way to address the 50 percent limitation issue when a network is fractured into a multiple discrete environment. This is a way to help an organization that may be growing one part of its network faster than others to get space it needs to meet this need. It also addresses the possibility of gaming the system by requiring organizations to reallocate space to its growing network first, which we hope will prevent hoarding. As a point of clarification, Bill added that this policy proposal was the result of ARIN staff bringing the issue to the attention of the Advisory Council.

Comments:

  • General Comments:
    • My company has a network with 40 locations, each appearing independently and not interconnected. Right now they mostly have /28s and /29s, most of which are 80 percent utilized. If I decide that I want to advertise things that will make it possible to filter from each of those locations, under this proposal it appears that I'm immediately entitled to a /15, 40 times the addresses in a /20. The authors may want to think about if this is the kind of loophole they want to introduce.
    • The notion that you need a certain sized network to get past filters is something I don't personally believe is true anymore. I'm particularly concerned with the idea of a /20 boundary because ARIN is making allocations down to smaller boundaries and I hope people are honoring those on the network.
    • The policy, as it currently exists, states that an organization must not allocate a block larger than the current minimum allocation size of the RIR and in parenthesis says that is currently a /20. The proposal says outright a /20, so perhaps we should revert to the language specified in the original policy.
  • Statements For and Against:
    • Andrew Dul stated that he was the author of the current policy on multiple discrete networks and that he supported this proposal. The problem it is trying to solve is that for very small networks, it is hard to get to 50 percent at the outset, so this provides for an exception that you just can't have any /20s available.
    • My small company has multiple discrete networks using a /20 at each location. Our last site was LA and it was not growing as quickly as other locations. If we need to roll out another site, the LA location would probably not be at the 50 percent utilization yet. This proposal would allow us to roll out the next site and still be able to grow the other networks at their current pace. I'm in support of this proposal.
    • This policy would definitely help my organization. We roll out new products that, due to regulatory restrictions in some cases, cannot interconnect without other networks. We would like to have the option of assigning the /20s to each individual market. We're rolling out products at a high rate, and we often don't end up with much immediate utilization.
    • My company could have used this a few years ago and would have benefited greatly. And while we don't need this today, I can see where this would help others as they get off the ground or are growing.
    • This wouldn't benefit my company at this moment, but I could see how it would benefit others and I totally support this proposal.
  • Questions/Responses/Clarifications:
    • For ISPs, the utilization requirement for additional space is generally 80 percent and I see a 50 percent requirement in this proposal, but it also says 80 percent for any given location. Is the intent still to require ISPs to achieve an 80 percent overall utilization, and it's just that when you have multiple discrete networks, you can't accomplish that? Andrew Dul responded that the intent of the current policy was for organizations with networks that were not connected yet to announce a block of some size in order for it to be globally routable, i.e. a /20. As their networks grow, they grow at different rates. Before the current multiple discrete network policy, you couldn't get more space unless you had the 80 percent overall utilization, and that's why the current policy specified 50 percent. It was just as random a choice as 80 percent. The network that would be receiving the additional space would still need to meet the 80 percent requirement, but overall the organization could have 50 percent. In this new proposal, an exception has been added that if you don't have any /20s available, you're still allowed to get another allocation, regardless of whether or not you've met the 50 percent overall utilization.
    • A concern was voiced about the system this proposal would establish, and its ability to be gamed by ISPs to request space under this policy without meeting utilization requirements. ARIN staff was asked to comment. David Huberman, ARIN Senior IP Analyst, answered that this policy was attempting to address a simple math problem. Current policy is biased against small networks in that large networks can overcome the 50 percent utilization across their last allocation fairly easily, and have lots of POPs or lots of address space. Small networks are forced to find strange and sometimes bad ways to get around the policy, so this proposal is trying to address that issue.

Polling of consensus:

Question: Approve of Policy 2004-5?
Yes? 55 No? 0

Policy Proposal 2004-3: Global Addresses for Private Network Inter-connectivity

Introduction: Einar Bohlin, ARIN Policy Analyst

Presentation (Read-only): PDF PPT
  • Policy Proposal Introduced - March 17, 2004
  • Presented at ARIN XIII
  • Revised text - September 20, 2004
  • Staff impact analysis - October 4, 2004
  • Legal review - October 7, 2004
  • PPML Summary:
    • No discussion since revised text was posted to PPML.

Presenter: Marla Azinger, author

Presentation (Read-only): PDF PPT

Marla started the discussion of the proposal with an explanation of the work done so far. This proposal is attempting to clarify policy for something that is currently done, but not formally addressed by ARIN policy. This policy was reviewed and voted on at ARIN XIII, receiving a majority vote to make this its own policy, pending some rewriting. Through meeting with the AC and using the input from the past ARIN meeting and comments from PPML, a new version was created. She then read the new proposal text and asked for comments.

Comments:

  • General Comments :
    • I would recommend we go forward with this only in the context of IPv4.
    • I believe the issues that caused this proposal to be written will happen in IPv6, so I think a general policy is applicable. However, because the IPv6 community has not fully defined things like public and private IP address space, I believe this policy should be IPv4 only for now. Once a definition in IPv6 is done, we should adopt a similar policy for that protocol with the appropriate terms.
    • The proposal should say something about ARIN not requesting name server information for these networks or we're going to end up with more lame delegations. It should be clear that just because someone gets an allocation for a private disconnected network, there can be no expectation that DNS will work as usual.
    • I think it would be a bad idea to refuse registration of name server information if that is what people want.
    • Since people can already ask for an allocation and not have to submit name server information, I see no reason to alter this proposal to make that an option.
    • As part of educational documents, rather than as part of the policy, people should be encouraged not to setup the potential for lame delegations.
    • Since there is nothing in the proposal that would stop people from using space that was initially intended for internal use from being used externally, there should still be the option for providing name server information.
    • I would prefer if these networks were issued out of a reserved block that would be known not to be globally routable.
    • Remote participant statement: I would like these networks to be marked in WHOIS as private networks and the allocations to be made from a reserved block.
    • While we were talking about this, I attempted to rewrite it without reference to the RFCs and make it IPv4/IPv6 neutral. It would read as follows: "Networks not presently interconnected to the global Internet may obtain IPv4 or IPv6 address space from ARIN if they will be connected to the global Internet in the near future or if use of private IP addresses would produce imminent addressing conflict."
    • If space has been marked as globally non-routable, then the ISPs can ensure these ranges are not used by spammers by choosing whether or not to accept these announcements.
    • The issue of providing space so that interconnected networks can operate without conflict is a separate issue from networks that are private, but may in the future be announced publicly. No one should be able to get public address space for the latter situation; they should use RFC 1918 space now and renumber when they get public space. We need to separate these issues.
    • The issues of network addressing conflicts when interconnecting private networks could possibly be addressed with the expansion of RFC 1918 space.
  • Statements For and Against:
    • This policy is asking us to do something very inefficient, in that it allows ARIN to assign global addresses for use as private address space. This does not seem in keeping with ARIN's mission of stewardship. If they need more private address space, they should use more RFC 1918 space.
    • I'm in favor of the policy and I just have one comment about the question of public/private identification in the database and having these come from a reserved block: I'm against all that because I think it opens up these blocks to abuse by spammers.
    • I am not in favor of this because of the way it encourages the use of global addresses for private networks.
  • Questions/Responses/Clarifications:
    • Does there need to be an explicit recognition that this proposal would apply to both IPv4 and IPv6? Marla replied that she had received feedback that someone wanted to go forward with this for IPv4, but nothing clear about IPv6. She also explained that she had received information about something like this for IPv6 in one of the other regions, and perhaps the AC could address this to make it clearer. In clarification, an attendee added that one of the reasons this might have been left unclear for IPv6 is that the idea of private space in IPv6 is still evolving and until an RFC is completed, we don't have any guidance to point at.
    • Do we need the proposal to mention RFC 1918 space, or can it be simplified to refer to private space and the appropriate use of such space?
    • Does this proposal imply that ARIN should provide network topologies, private network information, and information about assignment strategies to verify the need to avoid network conflicts? Marla replied that verification requirements already in existence would still apply.
    • Is the intention to have a separate block for these allocations to be made from, so that they can be filtered on? The recipients of these networks could start announcing it to the public six months after the allocation. Marla replied that this was discussed at the last meeting and the consensus seemed to be that they would not be from a separate block.
    • Does current policy include the requirement in the proposal that organizations prove they've used RFC 1918 space efficiently before they can get global addresses? An attendee responded that ARIN shouldn't be encouraging the use of RFC 1918 space. It can acquiesce to it, but it is not useful for ARIN to encourage the use of NATs, etc.
    • Does ARIN have any mechanism for generating some kind of daily text file that shows all address space that is known to be private? People could then generate a filter on that so that space could be black holed. John Curran responded that there is currently no such marking and ARIN does not generate such a list. If there were such a marking, the data could be disseminated in many formats.
    • John Curran asked if the purpose of this proposal was to allow organizations to get public address space for private purposes that eventually may connect to the Internet, or if it is meant to address the problems organizations have in interconnecting private hosts. Marla replied that the consensus expressed at the last meeting was for the policy to address both. As a follow-up, John asked if this was something that could be addressed through better education about current policy or something that required a new policy. Marla explained that this also was discussed at the last meeting, and while it was agreed that education was important, a consensus vote supported having a clear policy on the issue.
    • There was discussion of the impact of the new NRPM related to this, and that perhaps a new policy was not needed, but instead a general clarification could be added to the document to explain what the policy was. Lee Howard responded that adding anything to the NRPM would require a policy proposal, as the entire document is "policy." It's not a question of a minor edit. Einar Bohlin added that everything in the NRPM is policy.
    • Will the global address space that is given for use in private networks be considered in terms of justification and utilization when coming back to ARIN for more space? Marla responded that the proposal provided no exemption, so the rules about utilization would still apply when requesting additional addresses.

End of Discussion:

John Curran, as moderator, judged that there was not consensus to take a formal sense of the room and that the AC would take the feedback provided today to determine how to move forward.

Policy Proposal 2004-8: Allocation of IPv6 Address Space by the Internet Assigned Numbers Authority (IANA) Policy to Regional Internet Registries

Introduction: Einar Bohlin, ARIN Policy Analyst

Presentation (Read-only): PDF PPT

Einar introduced Scott Bradner to speak about the ARIN Internet Policy Evaluation Process and its application to this global policy proposal.

Scott explained that we have two types of global policies. One is the kind where all the RIRs agree on doing the same thing, like we did with the global IPv6 allocation policy. The other, and what we'll be discussing now, is that the RIRs will not implement the policy themselves, but will instead rely on an outside body to implement it. In this case, this is a policy for ICANN's management of the IANA. We'll be following the process outlined in the IRPEP up to, but not including, ARIN staff implementation. Once the proposal is ratified by the ARIN Board, it will be passed up the chain to ICANN. There it will undergo their review cycle and either be accepted or passed back down to go through the whole process again.

Einar continued with the proposal introduction.

  • Template submitted - August 19, 2004
  • Posted to PPML - September 20, 2004
  • Revised text - September 22, 2004
  • Staff impact analysis - October 4, 2004
  • Legal review - October 7, 2004
  • PPML Summary:
    • One post in favor

Presenter: John Sweeting, ARIN AC

John began by covering some of the background of this proposal, noting that it has been or will be discussed by each of the RIR communities. The text is nearly identical to versions under discussion in the APNIC region and at the RIPE NCC and LACNIC meetings. The only difference is the language describing the allocation term to the RIRs, in that a three-year term was discussed at APNIC and in this proposal it is 18 months. He then briefly outlined the allocation principles described by the proposal.

Comments:

  • General Comments :
    • Instead of making this policy, ARIN needs some sort of recommendation process that is an abbreviated form of the policy proposal process.
    • In section 4 concerning the announcements IANA would make upon the release of an allocation, I suggest that we give IANA more flexibility.
    • The current text mentions establishing the need for an 18-month period, but does not mention how to establish the basis for that. Also, it says that a 50 percent utilization is required before asking IANA for more space. It could happen that an RIR with less than 50 percent utilized could have enough space to serve its community. These two triggers are not necessarily compatible. I suggest something more like what was approved for the IPv4 version of this policy. The requirement should be based on average utilization extrapolated forward.
    • The requirement should be based on projections established by real operational data, like current burn rates.
    • Doug Barton of IANA commented that Section 2 of the proposal says the initial allocation should be a /12 and then Section 3 says subsequent allocations should be a /12 or multiples of a /12. I think having these limits be the same is probably the wrong direction to take. Having a smaller subsequent allocation would give IANA better flexibility going forward. Rather than a specific size, the best strategy might be to do it in 4-bit boundaries, so whatever the community agrees is a reasonable initial allocation, the subsequent allocation unit would be the 4-bit boundary below that. Also, it should be noted that announcements of allocations from IANA are only going out to mailing lists where they have been requested. It would be useful to amend this policy to say IANA will send announcements in the format specified, limited to what the block is and to whom it was issued. IANA can work out with the RIRs the timing of how these announcements are done to address some of the concerns that have been raised. In addition, it would be useful to have guidelines on how to evaluate what 18 months of space means.
    • I suggest that the AC break apart the issue of the announcements from the rest of this and focus simply on the technical elements that need to be in there. Doug Barton added that ICANN would support that amendment.
  • Statements For and Against:
    • This is a very important policy to move forward in the general sense, and it has important implications for how well we will be able to aggregate addresses in IPv6. It is important to get a global understanding of the issues involved.
    • It is important that something is done in this area as there is a risk in not doing anything. If the current policy of handing out /23s continues, it will create a huge mess. In fact, a /12 might be considered conservative. I don't think we can wait for it to be perfect and I support it as-is to avoid creating another swamp space.
  • Questions/Responses/Clarifications:
    • If this is not a policy ARIN would implement, why will it be included in the NRPM? Ray Plzak responded that as it is a policy that effects how ARIN does business, it needs to be reflected in its policy documentation. Scott Bradner expressed that he did not believe it should go into the same policy document. John Curran added that this would be something for the Board to address.
    • Why do we need a policy for this any more than we need a policy on how ARIN deals with its landlord or electrical utility? Scott Bradner replied that it is establishing a policy for ICANN, where the registries can suggest a policy to ICANN and they can evaluate it and decide what they want to do about it. It is reasonable to do that on a global basis, whereas it would not be reasonable for them to buy electricity on a global basis.
    • What are the criteria for deciding when an RIR needs another allocation? This is important because if you try and get high utilization, the RIRs won't go back as often, and you won't get good aggregation. Ray Plzak answered that the policy does not dictate how IANA exercises its reservation policy and we can only hope and assume they would take that into consideration. Currently they are not doing that with the allocation of /23s, and as a result we have a shotgun effect across the RIRs. Doug Barton added that he agreed this was an important issue and various initial versions of the policy proposal in different regions either did or did not specify globally reserved prefixes for each region. The matter has received considerable debate in the community as to whether those reservations are a good idea. IANA is open to input on this. There should be a chance to address this in the review cycles of this proposal before it reaches the final approval in the ASO/ICANN process.
    • Can someone speak to why the /12 prefix length was chosen?
    • In response to a question about this proposal's status in the other regions, the following responses were given. Son Tran, of APNIC, said that it had been approved there. Axel Pawlik, of RIPE NCC, said it was approved at their Manchester meeting and was now being discussed on the mailing lists. Raúl Echeberría, of LACNIC, explained that a formal proposal on this had not yet been made in their region, but it will be discussed at their meeting next week in Costa Rica and they expect to have something at the meeting after that.

End of Discussion:

Discussion occurred on whether to divide the discussion and remove Section 4 of the proposal from consideration. Additional suggestions on other text changes were then discussed. It was decided that lacking a clear consensus on the nature of the changes that were under consideration, the proposal be remanded back to the Advisory Council for additional work.

Lea Roberts of the AC requested a poll to judge the sense of the room regarding raising the minimum allocation size above a /23, which is what is currently in place.

Question: Should the minimum allocation size from the IANA to the RIRs be raised significantly from a /23?
Yes? 33 No? 0

IETF IPv6 Addressing Update

Presenter: Thomas Narten, ARIN IPv6 Working Group Chair

Presentation (Read-only): PDF

Thomas presented an overview of activities in the IETF on IPv6 addressing issues, specifically globally unique local addresses.

Highlights:

  • RFC 3879 has deprecated IPv6 site local addresses.
  • Recommendation to define a range of addresses intended for intra-site communication. They would be globally unique, not intended to be globally routed though would be routable between consenting parties.
  • Globally unique prefix for each site to eliminate address collisions when sites interconnect.
  • ISP independent, but are not “PI” addresses.
  • Two types of IDs: one would be locally administered and rely on 40-bit random number generation; the other would be centrally administered, by an as yet unidentified authority, for those who wish to guarantee uniqueness.
  • Centrally administered IDs would be permanent, with no fees associated, and could not be reclaimed.
  • For centralized allocation, the IPv6 Working Group is finalizing some details and there are continuing discussions between IETF, the NRO, and the RIRs. Local allocation is currently under IESG review, and should be approved soon.
  • IPv6 ad hoc Advisory Committee has been created by IAB to provide advice to IAB on IPv6 addressing and related issues.
  • Additional work items include issues of IANA to RIR allocations, review of HD-Ratio, and the deprecation of ip6.int.

Comments:

  • Concern was expressed over the varying quality of random number generation and what impact that will have on network conflicts. I also think it is overly optimistic to think that just because people say they won't route these addresses, that will actually be the case. What would be the business case to route or not route them? It seems there would be a huge financial incentive to route them, and they would end up in the global routing table. Thomas replied that he believed people would route these locally, but they would not end up in the global tables. Because of how IPv6 works, they can use two addresses. One that is globally routed to connect to the Internet, and another globally unique local address to connect to a private network.
  • I agree that given financial and other types of pressure, companies will try to route these, and that would pollute the routing space. People want provider independent space, and simply saying that isn't what it is intended to be will not stop people from using it that way. I think this is a stunningly bad idea.
  • I also agree that both the faith in the quality of the random numbers used and the belief that these will not end up being routed are naive. I'm also concerned about the definition of permanent. How will we track who the owner, or whatever term you want to use, is? How do you enforce that the space is being used by the correct person or organization? What mechanisms would be in place to prevent hoarding? I agree that ownership should be kept private, but how then does anyone verify who is using it?
  • This is a bad idea from many perspectives. The idea of global routability is a fiction. Also, as far as "can't be taken back," that means it's my number, my estate's number, and my assigned heirs' in perpetuity. Thomas responded that it might sound like a crazy idea, but people have done the math and are comfortable with the fact there is never going to be a need to reclaim these.
  • While there are a lot of IPv6 addresses, it is not infinite. It might take 100 million years, but they'll run out eventually especially if they're permanent as that will mean we'll go through them pretty fast.

aflow – a Netflow Packet Analyzer

Presenter: Dan Ardelean

Presentation (Read-only): PDF PPT

Dan, a former ARIN intern, gave a presentation on a tool he created for the analysis of net flows.

Highlights:

  • ARIN has agreed to release this as an open-source tool under GPL license.
  • Major goals were the ability to easily compile and install the tool, a powerful configuration capability, and the ability to do flow analysis based on filters and classifications.
  • Features real-time capture and classification of net flows using a simple and powerful configuration language.
  • Extendable through modules, so functionality is easy to add.
  • For announcements and downloading, contact aflow@arin.net.

Open Microphone

Moderator: John Curran

John Curran opened an Open Microphone session for comments.

  • How do we create a process for recommendations to address issues that are unsuitable for the policy evaluation process? John Curran answered that, based on feedback from this meeting, a group within ARIN would be working on that. Scott Bradner added that he would draft something.
  • Andrew Dul, of the ARIN AC, raised the issue of near-real-time publication of net blocks, and while this was discussed at a previous meeting and Leo Bicknell had presented the topic to NANOG, the AC would like some feedback. There was no clear consensus at the NANOG meeting on how this should be done, though suggestions included BGP feeds and RPSL publishing. If there is no clear direction from the community, we might just need to drop this. However, I think that would not be the best thing to do. Lee Howard responded that we've heard this come up a couple of times today, and he thought it was still something that needed to be addressed. Leo Bicknell added that the response from NANOG was surprisingly lukewarm and that the people who responded were less concerned about the real-time aspect of it. They wanted it up to date, but their focus seemed to be on it being provided in a number of formats.
  • Scott Bradner again raised the issue of a recommendations procedure to get feedback on ways to reach consensus in an abbreviated fashion, specifically the use of mailing lists. An attendee suggested that to raise awareness of discussions involving recommendations, a post to arin-announce@arin.net that pointed people to the PPML discussion would be helpful. Another attendee suggested that the current online voting tools could perhaps be adapted so that people who did not wish to express their opinion on a public list could still participate. Another suggestion focused on having two stages of recommendations, one that was weaker that would involve a quick poll on the mailing list, and another stronger one that had come before the membership at a meeting. Scott Bradner suggested the following process: a "design" team of some kind would come up with a proposed recommendation, put it up for discussion on the mailing list, possibly revising it in the process, and then send it to the AC or a sub-group of the AC where it would undergo possible further review and revision. He thanked everyone for their input and said he would draft something to pass along to the AC and the mailing list.
  • Lee Howard asked for clarification on the judgment of consensus regarding proposal 2004-3, adding that he wasn't sure there was sufficient guidance to the AC. He invited anyone to speak who wished to voice support for the proposal as written or lack of support unless a particular item were stipulated. An attendee asked ARIN staff to explain current practice regarding requests for space that is specifically not going to be connected to the Internet. Leslie Nobile, ARIN Director of Registration Services, responded that in certain circumstances they are given space, but ARIN generally discourages the practice, which is the guidance from existing policy. We do ask for very thorough technical justification as to why they need global addresses for use on private networks. A follow-up question was asked about any commonalities that existed in those organizations who had successfully requested this type of allocation. Leslie responded that there was no commonality, though financial networks are often seen. Lee added that the current policy's wording does not outline any procedures for organizations that resist the encouragement to use private address space and that the policy needs to be explicit one way or the other. An attendee asked if the allocation of global space for private networks was considered when an organization that already received one came in for additional space. Leslie responded that it was treated like anything else in that it had to meet utilization requirements and ARIN had to be able to see the utilization. Rob Seastrom, of the ARIN AC, asked if ARIN staff could provide a scope for the issue in terms of the amount of address space we're talking about per year. Leslie replied that there were 7 requests in the last year, with some of them for large networks of /9s, /10s, and /11s. Those actually granted were between /16 and /18.

Upon additional discussion, polling of the consensus of those in attendance was done on the following issues related to proposal 2004-3:

  • Strike requirement to have used RFC 1918: For 6, Against 6
  • Use of globally unique space:
    • for interconnects with other private networks: For 32, Against 4
    • if requester is out of 1918 space: For 16, Against 12
    • if requester may connect to public network in future: For 16, Against 12
  • Support for proposal as is: For 15, Against 19

Closing Announcements

Ray Plzak made closing announcements, including a reminder for attendees who were not staying for the Members Meeting to turn in their wireless cards and to fill out the meeting survey. Ray again expressed thanks to the meeting sponsors: America Online and Time Warner Cable. He reminded the audience that the Members Meeting would begin the next day at 9:00 AM EDT.

Adjournment

The meeting was adjourned at 5:19 PM EDT.