The train it won’t stop going— no way to slow down.
--Jethro Tull, Locomotive Breath
Based on the rates at which the IANA is allocating IPv4 address blocks, as discussed in Monday’s post, it will run out of IPv4 addresses sometime in 2010. Because the RIRs maintain an 18-month supply, their pools will run dry 12 to 18 months later. How long the LIRs and individual end-users hold out after than depends on how many addresses they have in reserve when the RIRs can no longer allocate.
Saying there is less than a four-year supply of IPv4 addresses remaining sounds alarmist, and there are still plenty of people around who will make just that accusation. But the numbers are there for anyone to observe and make their own conclusions. What’s more, there are factors beyond just the numbers indicating that, in fact, new IPv4 address assignments are likely to become unavailable much sooner.
Various jokes have been made to the effect of, “Can you imagine the justification you would need to acquire the very last IPv4 address?” We cannot afford to let the IPv4 pool run completely dry; some number of address blocks must to be held in reserve for critical needs.
A proposal was made in March of this year to do just that. The proposal called for the IANA to hold 10 /8s in reserve, essentially stopping allocation at that point. The day allocations stop would be the “T” (Termination) date; a warning—sort of a two-minute whistle—would be issued on an “A” (Announcement) date, which would be when the IANA pool reaches 30 /8s. The assumption is that the T date can be accurately predicted within 2 years of the A date.
After discussion ARIN abandoned the proposal in April, when consensus could not be reached. (The proposal is still under discussion by APNIC, but is unlikely to reach consensus there either.) Considering that at present rates the A date would be reached sometime next year, this might have seemed too aggressive for some. But it’s important to note that the idea itself was not rejected, and holding some number of /8s in reserve is still likely to happen—shortening the time left for IPv4 allocations by, probably, one year.
There’s another factor that will shorten the time left for IPv4: A “run on the bank.” As more and more organizations realize that they will not be able to get new allocations when they need them, they are likely to increase their immediate allocation requests, perhaps faking the data to make their short-term needs appear more severe than they are. If such a phenomenon arises, it should be apparent within the next 12 months.
So is there any way to slow IPv4 depletion? A few solutions have been put forth. The most commonly heard suggestion is that the RIRs or IANA should institute a reclamation project, reacquiring unused IPv4 space either by economic incentive or by legal challenge. But neither of these is cost effective. A buyback program would certainly kick off a price surge similar to what a black market would do, making any IPv4 address expensive and a /8 a truly precious commodity.
Legal approaches could be even more expensive and impractical. Forcing the return of a /8 by fiat would likely mean a year or more in court, for the gain of less than a month of additional IPv4 allotments.
The IANA could enact more stringent rules for the remaining allocations, or perhaps add a monetary cost to them, but this would meet with loud political resistance, particularly from developing nations: “North America, Europe, and northeast Asia have had all the IPv4 addresses they needed for decades, and now that we’re building infrastructure you say the rules are different for us? I don’t think so!”
The IANA could open up the largely unused Class E address space (240.0.0.0 through 255.255.255.254), originally reserved for experimental use. This would add 16 /8s and around a year of additional allocations. But because these addresses have always been reserved, most host and router operating systems are coded to view addresses in this range as invalid. The investment required to upgrade code again far outweighs the short-term benefit.
The IANA could use one or two of the remaining /8s to expand the RFC1918 private address space, allowing much larger topologies behind NATs without having to build layered NAT hierarchies. But this just exacerbates existing NAT-based network and application limitations, does nothing to alleviate the growing demand for globally unique addresses, and would again require upgrades of some software to recognize additional private IPv4 addresses.
Interestingly, there is one factor that could indeed slow the depletion of IPv4 addresses: The increased deployment of IPv6. As more and more IPv6 networks and services are built, the need for IPv4 addresses decreases. In an ideal world filled with sunflowers and kitty cats, IPv6 implementers might even return IPv4 address blocks they no longer need. Reality, unfortunately, is going to be much messier over the next five or six years as network operators are forced into some hard decisions in the face of relentless growth of network infrastructure and services.
No way to slow down.
I’ve been evangelizing IPv6 for some eight years now, and until very recently the response I faced most often, at least in the western hemisphere, is, “How do we make money with it? How do we get a return on investment? Show us the business case.”
In the next post I’ll talk about the IPv6 business case.
Jeff Doyle is president of Jeff Doyle and Associates, an IP network consultancy. Jeff is the author of Routing TCP/IP, Volumes I (read an excerpt) and II and of OSPF and IS-IS: Choosing an IGP for Large-Scale Networks. He is a frequent speaker on IPv6, MPLS, and large-scale routing.