IPv4 Runs Low Faster Than IPv6 Is Adopted
Posted by David Greenfield on January 5, 2010
On the Seinfeld episode, "The Dealership," Kramer takes a test drive. The salesman asks about gas and Kramer responds, "There's still some overlap between the needle and the slash below the "E"...I've been in the slash many times. This is nothing. You'll get used to it."
According to engineers at Hurricane Electric and iNetcore, by mid September, 2011, the Internet registries are expected to run out of routable IP addresses to assign. We'll be approaching "the slash" soon enough as addresses for users, hosts and devices will suddenly become a scarce resource. While 2012 apocalyptic visions aren't expected, the problem remains serious. Migrating to IPv6 is the leading approach, with its much larger address space of a little more than 340 trillion addresses. IPv6 could also introduce a number of engineering headaches for enterprises and consumers.
During the early days of the Internet, publicly routable addresses were awarded liberally. As such, organizations involved in the building the Internet accumulated massive blocks of IPv4 addresses. MIT and IBM have some 16 million IPv4 addresses allocated to them. AT&T holds twice that many. Neither of those organizations are likely to run out of addresses any time soon. While everyone agrees that IPv6 may provide the long term answer, how to get to IPv6 is a different matter. Until that's worked out, a number of provisional approaches are being considered to ease the demand for IPv6.
There have been proposals to reclaim some of unused addresses and sell them on the open market. The registries who assign addresses to ISPs would manage this marketplace. Such an approach will only create more complexity and delay the inevitable by about a year, says Leo Vegoda, the number resources manager for ICANN, and the person responsible for looking after the IP address and autonomous system numbering spaces. Loosely defined, an autonomous system refers to a set of IP subnets and the organization that manages them, that collectively route to the Internet. Autonomous Systems (AS) are assigned Autonomous System Numbers (ASN).
Determining the addresses that can be reclaimed from an AS is no simple feat. "The fact that an address is not routed doesn't mean that it's not being used," Vegoda says "Nor does it mean that they are using the whole allocation. It's just not clear." However, in the bigger scheme of things a reclaiming those addresses wouldn't help all that much "We allocate a /8 per month, roughly," says Vegoda, "Each /8 is 16 million addresses. If they could reclaim a half dozen or dozen /8s, that would only push out the issue by a year."




Comment by Smithwill on January 5, 2010 10:06 AM
I heartened to learn that AT&T is fully IPv6 compliant. This obviously explains why their service is so unreliable and spotty. It's not their fault! Because everyone else has been so slow to adopt the new address scheme, they're only able to speak to themselves. So, knowing this, I'm inclined towards giving them a break....until 2012 when everyone else catches up. Just kidding.
IP addresses, like carbon credits, will become a hot commodity to be bartered, bought and sold by the market makers. Consumers, once again, will take what they're given and be glad for the opportunity. Buyers of Internet bandwidth and services should begin to pucker up in preparation for the inevitable fork-lift upgrade to make their systems compatible. However, this won't be a problem for Cisco customers as they've been conditioned to fork-lifting their entire infrastructure about every three years whether they need to or not. Cisco really does care for their customers because they'll all be ready when the sh..IP (shortage of IP addresses) hits the fan.
Reply to this comment
Comment by Todd on January 8, 2010 9:37 PM
Typo: ipv6 usable address space is not 340 trillion ip addresses. It's much MUCH bigger than that: 34 x 10 to the 38th power, or 340 trillion trillion trillion
-- courtesy of Sean Michael Kerner of www.enterprisenetworkingplanet.com
Reply to this comment