Standards Road Is Long, Winding... Bumpy
To the chagrin of the IETF, the absence of IPv4 also pushed translation back into the spotlight. NAT64 translation of NAT64/DNS64 fame, works, at least most of the time. For the cases where it doesn’t work (Skype, Spotify, Netflix on Android) T-Mobile US Inc. employed even more translation -- in fact, double translation with 464XLAT. As explained by Cameron Byrne, director of technology at T-Mobile US, “A lot of folks in IETF, myself included, thought double translation, what an abomination. But at the end of the day it solved a problem.”
Regardless of criticisms about its impurity, it works; and T-Mobile is building out its IPv6 network of the future… and its 5 million+ subscribers on IPv6 haven’t noticed a thing.
Standards are important, but the customers, who in this case are the large Internet service providers, are more important. The tunneling standardized in Hong Kong and the transition mechanisms before it were technically excellent and made logical sense, but they didn’t have buy-in from those footing the bill: the customers. When that much money is at stake, it becomes bigger than a technical problem and requires all stakeholders to be at the table.
So the premise of having a transition period, upon which the original transition mechanisms are dependent, was ultimately flawed. The customer didn’t want to transition to IPv6 until absolutely necessary, and by then there were few v4 addresses available to go dual stack or use tunnels based on dual stack.
As a sign of the times, the new trend is to deploy v6-only networks and connect them to the v4 Internet. Reality is driving this movement, but there is also new thinking, as Suresh Krishnan, current chair of the Softwires Working Group, tells us, “People are thinking about v6 as being the main transport protocol and v4 as a service protocol that people can offer.” Hallelujah!
We now find ourselves in the awkward years of Internet plumbing, where kluges and duck tape are the norm. But there are solutions. Fast-forward eight years from Hong Kong, and the flux in transition standards has settled down while critical mass is being reached. Today’s transition mechanisms may not be as elegant as those originally standardized -- but they work, and if all goes to plan, we’ll be peeling them off in five to ten years.
For a more details, listen to full interviews with Hain, Krishnan, and Byrne on The IPv6 Show podcast on iTunes, and visit the gogoNET community for Transition Mechanisms whitepapers and videos.
— Bruce Sinclair, CEO, gogo6
