If you listen to the buzz, the day of ipv4 exhaustion is nigh and operators need to be migrating en masse to ipv6. The problem with the v6 evangelism is that it doesn't take into account the complex incentives at work in the adoption of ipv6. A recent allocation by ARIN of a /9 block to U.S.-based network operator Comcast highlights this.

Comcast has been one of a handful of network operators publicly proclaiming their commitment to the deployment of ipv6. ARIN has been praising them for this. However, a few days ago ARIN dropped an enormous allocation of ipv4 addresses in Comcast's lap: approximately 8.4 million. The Comcast allocation drives home the point, overlooked by so many, that v6 deployment does not, at this stage of the game, decrease the demand for IPv4 addresses. Comcast is one of the most advanced v6 deployers in the United States and yet here it is gobbling up a huge chunk of the remaining v4 space. We'll note with some irony that the largest delegation ever seen by some was completed while there is ongoing debate in ARIN and other RIRs over how to ration the last /8.

Whether Comcast is being favored over other network operators is impossible to say without more information. But there does appear to be a contradiction between ARIN's member-driven Number Resource Policy Manual, which stipulates that organizations can request up to one year's supply of IP addresses, and Comcast's last three years of declining growth in high-speed Internet subscribers, which shows increases of about 1 million subscribers per year. (Last year, Comcast reported 1 million net additional subscribers, down from 1.3M in 2008, and 1.7M in 2007). However, Comcast also stated several years ago that it had exhausted (RFC 1918) address space used for internal management of its network. Needless to say, an allocation of more than 8 million addresses suggests further need for clarity around the incentives to migrate to IPv6.

21 thoughts on “ARIN grants Comcast IPv4 “mega-allocation”

  1. Did Comcast threaten to sue the ISOC / IETF for bad advice on IPv6 ?
    Did the allocation shut them up?

  2. ARIN doesn't comment publicly regarding specific allocations to members due to privacy issues, but ARIN applies the community-developed Number
    Resource Policy Manual to all requests in an equitable manner. (Note – If you have any actual material information regarding number resource requests that may be based on inaccurate information, please notify us via this link https://www.arin.net/resources/fraud/index.html and we'll investigate.)
    While parties are actively testing and deploying IPv6 to handle long-term Internet growth, IPv4 is still being used today to connect new customers by almost all ISPs.  This makes perfect sense, since IPv4 should be used while it continues to be available.
    /John
    John Curran
    President and CEO
    ARIN

  3. Perhaps Comcast is assigning public v4 addresses to customers that they used RFC1918 addresses for previously? Since they are deploying v6, it makes perfect sense that they are dual-stacking with public addresses across their network, hence the “need” for a /9. MM is always going on about dual-stack needing v4 addresses, like it's a revelation.
    You guys need to stop hating and use the clues supplied by your own blog posts.

  4. McTim:
    I am afraid you need to stop hating us, and start being a bit more objective and critical about the issues facing the IPv6 transition.
    You say that I am “always going on about dual-stack needing v4 addresses, like it's a revelation.” Unfortunately, for people who read only the ipv6 propaganda issued by certain institutions, it is a revelation. Granted, it shouldn't be, but since you and others are constantly trying to avoid facing the implications of that fact, the point needs to be made. As noted in our IGF Vilnius blogs, there was an entire workshop on v6 and an entire main session which managed to avoid noting that fact, until I brought it up.
    A month and a half ago, when you responded to our IGF blog post, you tried to deny the fact, saying “implementing IPv6 at this stage does not necessarily require the use of additional ipv4 address resources…” Now that the world's leading ipv6 implementor has gobbled up a /9, you've changed your tune, just ever so slightly: “Since they are deploying v6, it makes perfect sense that they are dual-stacking with public addresses across their network.”
    And of course we understand (as you note it was right in the article) how the replacement of RFC 1918 space with public addresses might account for that. For you to question that is pretty churlish and unfair.
    So it's a case of a very dark pot calling a kettle black. Seems obvious to me that your statements are motivated by animosity toward certain people (i.e., us) and by a rather sad attempt to rationalize whatever ARIN and the fabled “intenret technical community” says and does.

  5. Milton,
    I don't hate any of you personally, I just intensely dislike the loaded language and innuendo that IGP consistently uses. Now, on to the facts.
    I am not trying to, nor have I ever tried to “avoid facing the implications” of dual stack using v4 addresses. It's just the way it is, no getting around it.
    The two quotes of mine that you have used are not incompatible. If an org wants to use their existing v4 space to dual-stack, they can….if they don't have enough v4, they need more (obviously). My tune hasn't changed at all, I keep pointing out where y'all hit the sour notes and try to keep you honest.
    There is no need to rationalize in this case, ARIN followed their own policies, and granted an allocation. You have been given the link to report fraud if you have any evidence. if not I would suggest that you keep the muckraking to yourself. It is unseemly, plus it encourages the trolls to even more outlandish heights of conspiracy theorism! This is particularly unhelpful IMHO.

  6. FACT: Comcast funds ARIN Chairman Paul Vixie and ICANN Board Member Suzanne Woolf
    http://twitpic.com/2zucvh
    http://twitpic.com/2zuec6
    http://www.isc.org/about/leadership
    http://www.isc.org/software/aftr
    “Developed in concert with Comcast, AFTR 1.0 is intended to ease the
    transition from IPv4 to IPv6 by allowing legacy IPv4 end sites such as
    home PCs to interact with IPv4 content providers and services over an
    IPv6 carrier infrastructure.”
    http://twitpic.com/2zuec6

  7. Your pics are not evidence of your statement. Please refrain of making these kind of accusations.

  8. McTim is right. Dual Stack does not necessarily require additional IPv4 addresses. By definition, it requires IPv4 addresses (dual stacking means having IPv4 and IPv6 on the same host(s)). However, for hosts that already have IPv4 addresses, all one needs to do is add IPv6 addresses. For hosts that don't have IPv4 addresses, or, in some circumstances where hosts have RFC-1918 IPv4 addresses, it may be necessary to assign IPv4 addresses in order to feasibly dual-stack a host.
    Comcast does a lot of things with IP addresses besides broadband internet access. Additionally, the assumption
    that the rate of address consumption is 1:1 with the rate
    of customer growth is absurd. Many customers consume more than one IP address. In addition, there are other pieces of infrastructure that need addresses.
    IPv4 is running out. Until it does, of course people will keep deploying it. At least Comcast has the good sense to be deploying IPv6 along side IPv4.
    I really don't understand what IGP seeks to accomplish by encouraging people to ignore the reality that IPv4 is running out, that deployment of IPv6 in parallel is the only way to a smooth transition, and, that we will be completely out of IPv4 addresses before that transition is complete.
    I think it is a disservice to the internet and I don't understand the motivation.

  9. Vinton Cerf has been placed on the ARIN Board of Directors.
    Vinton Cerf rarely, if ever, participates in ARIN meetings or mail lists.
    What happened to the IPv4 allocations from Vinton Cerf and Bernie Ebber's WorldCOM ?

  10. “IPv4 is running out. Until it does, of course people will keep deploying it.”
    IPv4 Addresses do not WEAR OUT.
    IPv4 Addresses are recycled and re-used.
    Poor ICANN (IANA) Management and ARIN Mis-Management of IPv4 Address Space has created Artificial Scarcity.
    Owen DeLong promotes IPv6 for HE.NET on numerous forums.

  11. Owen,
    I am not sure where you got the idea that IGP is encouraging people to ignore the reality that ipv4 is running out, but it's certainly incorrect. IGP generally, and I in particular, have emphasized growing v4 address scarcity in three distinct publications and numerous blog posts since 2007. Where we differ, I think, is in our approach to the end game. Unlike many others, we have emphasized the fact that the transition to ipv6 will take at least a decade if not more. and we have emphasized (as in this article) that deployment of ipv6 does not mean that demand for ipv4 is going down any time soon. As a result, we have favored a more liberal policy toward trading ipv4 blocks, because we will face years of growing demand for v4, and we have been critical of what we call “ipv6 evangelists” who downplay or ignore the compatibility and economic problems associated with the transition. Speaking for myself, I would like nothing better than to see ipv6 transition completed tomorrow, because of its implications for end to end and address abundance. But the more we learn about the transition the uglier it starts to look. those problems don't go away by pretending they don't exist, we have to understand them and develop appropriate policy responses.

  12. Why didn't Comcast go directly to ICANN like the big boys do ?
    That is clever how ARIN is used to move IANA Assets to Comcast. Are those U.S. Government Assets ?
    How is a /8 shown on Comcast's books ?

  13. Yes, and I post with my real name rather than hiding behind anonymity.
    True, IPv4 addresses don't wear out and can be (and are) recycled and re-used.
    However, the scarcity is not artificial. If anything, NAT has artificially reduced scarcity at some rather extreme costs to innovation, flexibility, and user self-determination.
    However, even ignoring that issue, the reality is that the most optimistic projections for reclamation say a 5 year effort might yield as much as the equivalent of about 15 /8s. Since we've already burned through 18 /8s this year, we'd still be short even if those most optimistic projections were to come true.
    I find it especially amusing that you blame ARIN mismanagement in light of the fact that the vast majority of that allegedly reclaimable space is in legacy allocations which were, by definition, done before ARIN even existed and in light of the fact that far more IPv4 addresses are being consumed outside of the ARIN region than inside these days.
    In short, the facts simply don't support your assertions.

  14. Among other places, yes, we differ there.
    First, nobody is denying that IPv6 transition will last more than a decade. It would be foolish to do so in light of the fact that it has been going on for more than a decade already. However, I don't think that it will take another 10 years from now for most of the transition to occur.
    I think at this point, the major obstacle to transition is inertia and that we'll see a very large force applied in opposition to that inertia next year. That force will come in the form of actual address runout. Will transition occur instantly as a result? Of course not. But I think you'll see more than critical mass having IPv6 deployed in less than 5 years and I think you'll see growth in IPv4 reduced significantly below IPv6 growth levels in less than 2 years.
    Yes, we also differ in that I do not believe that “he who has the most gold” or “auctioning addresses to the highest bidder” are the ideal approaches to stewardship of internet address resources.
    I think that everything that can feasibly be done to educate and inform people of the impending lack of IPv4 addresses has been done. I think there has been more than adequate warning time for organizations that choose to pay attention to adapt and deploy IPv6. Indeed, the company I work for has put significant resources behind efforts to inform even our competitors. For anyone who is still ignoring the message at this point, well, I look forward to serving their former customers.

  15. “McTim consults for African ISPs on IP resource issues. His most recent full time job was with the European IP address registry RIPE. He was also an ISOC WSIS Ambassador.”
    McTim is not in North America or the ARIN region.
    That is one of the flaws of Internet governance.
    Anyone can walk across borders (virtually) they can not cross in person. It is ironic that IPv6 will
    help the U.S. Government keep people OUT, while
    IPv4 continues to work on The Big Island.
    The Africa IPv4 Allocations can then be re-used in
    the .USA. McTim is happy with IPv6 and the ISOC.

  16. What's annoying is that large service providers must use real IPv4 addresses instead of RFC1918 space (not sure if this is the case with the new Comcast allocation, but there are numerous presentations out there from folks at Comcast describing the fact that they ran out of RFC1918 space in 2005). That seems very wasteful to me and does not do anybody any good.
    But it is however not Comcasts fault. There has been numerous attempt to extend the RFC1918 space by using the 240/4 blocks. The first known idea of this usage came in 1988. None of the ideas has gain any traction in IETF.
    http://www-mice.cs.ucl.ac.uk/multimedia/misc/tcp_ip/8813.mm.www/0146.html
    http://tools.ietf.org/html/draft-wilson-class-e-02
    http://tools.ietf.org/html/draft-fuller-240space-02
    I must say that it is very disappointing that none of the above mentioned draft made it into an RFC. AFAIK, this is not going to be discussed at the IEFT meeting in China either. Why is the community so against extending the lifetime of IPv4 with the 240/4? Is it just because some people built a career around IPv6?

  17. The fact that I (or you) can participate in multiple regional discussions regarding IP addressing policies is a feature, not a bug.
    FYI, the AfriNIC community is discussing a policy whereby v4 resources allocated to AfriNIC (from the last /8) will not be allowed to be used in any other region. With the caveat that “no more than 10% of these resources may be used outside of the AfriNIC region, and any use outside the AfriNIC region shall be solely in
    support of connectivity back to the AfriNIC region.”

  18. Some MSOs use public IPs for the CM's mgmt interface, and I believe Comcast is one of them. I wonder if they have a near-term plan to assign just IPv6 addresses to them, allowing them to move away from RFC-1918 addresses and free up public IPv4 space.

Comments are closed.