The alternative to v6 is NAT and v4. For about 99.99% of mobile users, it makes no difference whether they’re behind a NAT. Switching to v6 has costs, setting up a bunch of NATs (one per million customers, say) has costs, but there’s no reason at this point to think that one will be more expensive than the other. The NAT approach has the large advantage that with a NAT you can talk to all the v4 web sites, IMAP servers, and all the other stuff that mobile users want. With v6 they can only talk to other v6 sites unless the ISP provides a NAT v6->v4 gateway, so they’re going to use NAT either way.

John does this mean that you’re expecting the Internet services market to become permanently closed to new entrants once all IPv4 is distributed? If post-IPv4 allocation era service providers can only exchange traffic with other IPv6-accessible service domains (which currently represent no more than a fraction of 1% of the total), and they can only communicate with the vast, IPv4-only Internet via non-unique, local-only IPv4 addresses that are mapped to an incumbent’s pubic IPv4, then doesn’t that imply that new entrants have no viable business model—i.e., that their emergence and/or survival is, at best, an extremely low probability event? If not, what’s going to make the difference?

John does this mean that you’re expecting the Internet services market to become permanently closed to new entrants once all IPv4 is distributed?

Of course not.  It means they’ll have to buy v4 space just like they buy everything else they need to set up an Internet service.

At this point the interesting question is whether the RIRs will try to maintain the odd rule that you can only sell IP space if you wrap a shell company around it, or will admit that there’s a market and help make it orderly and transparent.

I’m not opposed in principle to v6, but to me its costs seem a lot higher and its benefits a lot lower than the IETF line claims.

Thanks for the clarification. Although phrased as a denial, it sounds like you’re actually agreeing with me. With the re/distribution of finite IPv4 addresses being a zero-sum game, new ISPs should have no problem entering the market, provided of course that they start out by buying the necessary (i.e., indispensable + non-substitutable) IPv4 addresses required to do so from an incumbent IPv4 owner/service operator. Given the fact that every other factor required to launch a new Internet service (bandwidth, CPU/processing capacity, memory/storage, human talent) is infinitely renewable/non-depletable, and the supplies of all but the last have already grown so phenomenally (with the price falling in tandem) thanks to Moore’s Law that they might literally be “too cheap to meter” in the not-too-distant future, I would suggest that IPv4 addresses are *nothing* like the other Internet service inputs.

Moreover, apart from the sole (common though not universal) example of last-mile facilities~bandwidth, can you think of any other Internet service input that is both indispensable/non-substitutable and also obtainable exclusively from an incumbent competitor? That certainly doesn’t sound like the markets for network hardware, or software, or talent, or facilities~capacity across any other segment of any Internet service delivery path other than the last mile… so it what sense, exactly, do you think that the market for IPv4 addresses would be similar to any other market for Internet service inputs?

can you think of any other Internet service input that is both indispensable/non-substitutable and also obtainable exclusively from an incumbent competitor?

Off the top of my head, real estate for the office, and (in the short run at least) staff, but people manage to find office space and staff anyway. IP space is indispensable, but it’s awfully simplistic to call it non-substitutable and to assume that everyone with allocated IP space treats each other as competitors and will hoard it.

There are ISPs that give every customer a globally routable v4 address (common in the US) and there are ISPs that put the whole ISP behind a NAT with handful of public addresses (common in Asia). Since IP space has been approximately free, the natural reaction has been to squander it, but that’s clearly going to change.  As a simple example, if the going rate for IP space were $1/address, how long do you think MIT would hold on to their /8, as opposed to renumbering to a /12 and selling off the other 15 /12’s for $1M each? 

How much of the allocated space is even announced? Half maybe? There’s a lot of space that could be shaken loose if there were some incentive to do so.  Sure, eventually the easy shaking will be over and we’ll start to see practical rather than nominal depletion, but that could easily take another decade, which is approximately forever in Internet time.  Heck, in another decade we might even have a majority of hosts with v6 ready software and working v6 DHCP.

Real estate for the office? I think this would be a good comparison to IP addresses if we lived in a one-dimensional world (or a world in which all locations are exactly equivalent/identical), and 100% of the real estate was owned by commercial real estate developers, and 100% of all aspiring entrepreneurs required office space upfront in order to launch their new enterprises. In a world like that, with a steady stream of aspiring new would-be entrepreneur office seekers contending for a fixed, unchanging amount of space, what are the chances that innovation would *not* be stifled by the market power of landlords to command any price for that bottleneck land resource? If you were a land holder, and you failed to extract the absolute maximum amount from your tenants (perhaps by simply making them tenant-serfs, and demanding x% of their operating revenues), then you and/or your “real estate” would quickly be acquired by a more “rational” landlord… That’s just how markets work, or at least how they would work in a world where real estate really was as indispensable and non-substitutable as IP addresses are to Internet service delivery.

If you think that IP network operators don’t regard each other primarily as potential competitors and/or revenue centers, and don’t maintain very steep “reservation prices” (and large and completely idle inventories) for services that they know to be very valuable, I would encourage you to think about peering and interconnection negotiations work, even between networks with “complementary” traffic patterns. Sure this hasn’t been true to date of IPv4 addresses, but that’s only because until very recently treating IP addressing this way has been both unnecessary (because of the remaining IPv4 reserves, plus the hope that IPv6 would become progressively more substitutable over time) as well as “illegal” (because of the official inalienability of IP number resources).

Your point about squandering is a good one, but it has no relevance except maybe to help explain who the IPv4 brokers are likely to be in the months and years to come. And your MIT example is also nice, but if the “going rate” for IPv4 really were $1/address today, but it was $0.50/address last week, would even MIT be likely to dump all of their reserves today? Why not wait until the price is $10 or $1,000 or $1,000,000 per /32? And if/when MIT does eventually sell, do you anticipate that it will operate as an RIR-like needs-oriented IPv4 dealer, and see that some of the resources make it to new entrants, or would the university be more likely to sell to the highest bidder—because that’s what you do in a market, or perhaps on the assumption that willingness/ability to pay top dollar is the best indication of “need”?

People have long wondered about how much of the allocated IPv4 address space that is not publicly/observably routed is actually idle and thus ready to transfer. I don’t have an opinion on the question, because I am certain that most is not held by institutions like MIT, and also that whatever amount there is will not be enough to support growth or innovation for long (if any time at all), and certainly not enough to last forever. Unfortunately, that’s probably how long it will have to last, because once IPv4 is converted to “real estate” per your analogy, then the land holders will have even less incentive to actively support a dramatic expansion of new “land resources” ala IPv6—unless that new territory provides them with (at least) all of the same market advantages that they enjoyed under the previous land-as-bottleneck arrangement. Without their positive support, the expansion is technically impossible, but once incentives have been re-engineered as you suggest, any hypothetical future expansion that actually occurs is unlikely to make any difference at all. Despite the absolute abundance of all of those other inputs (bandwidth, CPU, memory, human ingenuity, etc.), the pace of Internet growth and innovation will thereafter be absolutely determined by a selectively imposed, completely artificial, single resource-specific scarcity.

I would like nothing more than to to hear a persuasive, internally coherent argument for how and why this won’t happen, because I can’t think of a more tragic outcome. But I’ve been asking broadly for a long time, and I’m still waiting…

Zing—ten points for the snappy rejoinder!
Maybe nobody else will notice that your examples do nothing to support your original claim that the market for *Internet access* or *IP transit* will remain open to new entrants (or perhaps you’re expecting that incumbent service providers in those sectors will be good sports and stop providing their customers with unique public IPv4—or that the ability to provide customers with usable IP addresses will just cease to represent a huge competitive advantage). Perhaps nobody else will will notice that what you call “extreme worst case scenario” is nothing more than the *consistent* (as opposed to selective, convenient, self-serving) application of market logic to the overall situation created by the privatization of IP number resources.

Hopefully people *will* remember that it was exactly that kind of magical thinking a.k.a. “market faith-based policy making” that brought us the current fnancial crisis—and act accordingly.

Since I didn’t say any of that, I don’t see how I can argue about it.

But don’t forget to get your flu shot.

It’s more than just shortsighted to continue to push NAT as a solution.  It’s downright foolish, and it’s suggesting that bad engineering is the right solution.  NAT has always been nothing more than a work-around, a band-aid.  A good one yes, great even (see my earlier post giving more than a few kudos to NAT).  But it is a compromise that few engineers would seriously choose if IP addresses were plentiful.  And there are many drawbacks to NAT (again, see my post on this.)

As for an IP address trading market, this is a lot of speculation without a lot of substance.  The argument is usually made with utter disregard for the impracticality of it, seemingly suggesting that a market will just suddenly open up on eBay or Craig’s list.  Doubtful.  There are so many levels of impracticality to a trading market that I’m skeptical.  Perhaps there are some entities with idle blocks they could carve off quickly and sell, but there are many impediments to a trading market.  Worst case, it will be short lived and limited, another shortsighted band-aid that only postpones (briefly) the inevitable.  I’d love to hear someone put together a really solid analysis (beyond page 1 of Econ 101) of why there would be a trading market, how it would work and how those involved would get around the many hurdles they would face.

We can continue to treat symptoms or try to eradicate the illness.  Typically the latter is better.

With all that has happened in the last decade and all the economic issues we are still facing, haven’t we learned our lesson on being short sighted?

Verizon has posted specs for any LTE device that wants to use its network. They are mandating IPv6 support.

Some relevant quotes:

“The device shall support IPv6. The device may support IPv4. IPv6 and IPv4 support shall be per the 3GPP Release 8 Specifications (March 2009)”. (section 3.2.4.1)

and

“The device shall be assigned an IPv6 address whenever it attaches to the LTE network.” (section 3.2.4.2)

IPv4 support appears optional: “If the device supports IPv4, then the device shall be able to support simultaneous IPv6 and IPv4 sessions.” (section 3.2.4.4)

Verizon appears to be trying to conserve IPv4 addresses by disallowing long-term address leases: “If the device supports IPv4, the device shall request an IPv4 address if an application using the LTE bearer requests a data connection using an IPv4 address. Once the application is closed, the IPv4 address shall be released by the device”. (section 3.2.4.3)