If there was a single take-away message from the 2012 Australian IPv6 Summit it was “IPv6 is all over your networks, right now”. I’d add: “…and people are using it to get what they want, right now.”
Earlier this year I wrote Slow IPv4 Spurs IPv6 Alternative after discovering massive numbers of downloads of the gogo6 tunnel client from our site by people who were sick of slow YouTube delivery in Taiwan and chose to use Freenet IPv6 tunnels to get faster service. They didn’t have the slightest interest in why or how IPv6 could do this for them, or if tunnels were the ‘right way’ to get IPv6: they just wanted access to something, and grabbed whatever worked to get it.
(Tunnels ‘encapsulate’ packets of one type inside packets of a second type, to transit networks of the second type – a reliable technique used in many areas of telecoms for decades. To get IPv6 traffic over IPv4 networks, providers can offer dedicated encapsulating gogo6 servers. All the users have to do is set up the client software to access the server.)
Lately, our website logs are once again full of downloads of the gogo6 client, now from all over the world. Apparently, pirated copies of Microsoft’s new Windows 8 RTM version became available on the Internet, but they needed activation keys. An illegal Key Management Server on an IPv6 address was providing those keys. Once again, people were perfectly happy to set up an IPv6 tunnel to get to something they wanted.
Those IPv6 users didn’t realise they were supposed to define a transition plan, achieve corporate buy-in, test their networks, get training, update software and hardware, and gradually move to native dual-stack services. They didn’t know tunnels are unfashionable, and native services are deemed to be the only ‘right way’ of implementing IPv6.
Today, IPv6 tunneling is seen as no more than the poor cousin of native IPv6: but tomorrow, when management wants it now, tunneling may be the ideal quick and not-too-dirty means of getting vital services up and going.
And one day – within a single equipment cycle, according to Fred Baker – IPv4 will be departing stage-right faster than anyone can now imagine. Dual-stacking is (twice as) costly, and once transition has actually started, the financial imperative to move to IPv6-only services will be compelling.
And then there’ll be legacy IPv4 islands left everywhere, essential software that can’t be transitioned to IPv6. How will those IPv4 islands remain connected? Through good old tunneling. It’s the main technique currently available to support aging IPv4-only equipment on IPv6 networks – in this case, carrying the IPv4 traffic inside IPv6 packets.
But that’s in the distant future, isn’t it? Well, no. At the Summit, Ken Wilson from Opengear described how tunneling for legacy IPv4 equipment is already necessary today.