I, for one, am going to be using this as an excuse to slack off at work tonight. "Sorry I wasn't bothering to answer preteen's questions about fashion. The internet wasn't working. No, not my internet. The entire world's internet."
I may have been the crazy cisco guy, not sure. Geekery to follow. Either way, I've been pretty interested in this and here's my take: The only reason we're having a IPv6 day at all is because there's been a stalemate between the major internet application providers (yahoo, google, facebook, etc). They predict 1 in 20,000 users will not be able to get to every day IPv4 websites if they even enable IPv6. This is because there's a 1 in 20,000 chance the client incorrectly believes it has a IPv6 connection, but it goes nowhere. For most people this is a negligible number, but for these folks it's significant because the user gets the idea that "google is down, I will just use yahoo". So.. IPv6 day. They all agree to enable IPv6 at the same time, sorry Mr 1/20,000. As a side effect it raises the awareness, publicity, etc. Facebook, yahoo, google, all keep their users without playing prisoner's dilemma. A little history for those interested: -1996: IPv6 created. IPv5 must have been shitty, as it didn't make the cut. -1999: NAT created. You know how you always get a 10.X.X.X, 172.16-32.X.X or 192.168.X.X number? That's because everyone else does too, and you're actually sharing a single real address with other people. It's the only reason we haven't had to migrate to IPv6 before now. The IT industry has said we will run out of addresses 'this year' since around 2002 or so. -2000-2009: Not much. US government started pushing it really hard, lots of people came up with ways to get around the address depletion, but none of them really worked out. -2010: We more or less start to come to terms we can't beat the address depletion. -2011: The global address registry (IANA) runs out of addresses in February. IANA gives address to 5 continent-based registries (RIRs). These registries give addresses to end customers and service providers. Asia (APNIC) ran out of addresses in April. The only people with any remaining address in Asia are service provider and companies with unused addresses. Microsoft bought Nortel's addresses for $7.5 million, roughly $11 an address. The registry for North America is predicted to run out in fall time, so probably before the end of the year. They basically expect all the existing addresses that aren't assigned to be assigned by the end of the year. This will mean all new addresses will either be 1) bought or exchanged on a secondary market or 2) given a IPv6 address. Repercussions and Opinion: -It should be said at this point the addresses are free - you request them, and as long as you have business justification, you are given them. Most likely, people have been requesting more than needed, taking a 'run' at the addresses. -The registries are allowed to do these IP transfers, so it won't be so much of a 'black' market. -Historically - there hasn't been a money-backed business case for IPv6. Not only has the IT industry at large (including Cisco) said we will run out of addresses every year, but there hasn't been dollars behind it. The government realized it, because they move so slowly, to start 7 years ago. Most haven't. That's why I think the Nortel deal is so interesting. You're now sitting on money you weren't before. -This means most companies are probably sitting on $150,000-$1,000,000 worth of addresses ( 50 class C's to 1.5 class B's). -The whole basis of new IPv6 addresses being able to contact existing IPv4 addresses hinges upon doing a translation from 6 to 4. Because 6 isn't backwards compatible (thanks 1995), you have to do conversions. Which basically means we're doing a type of NAT again, which is one of the reasons we're trying to get away from v4 to begin with. The protocol-du-jour for this is stateful NAT64. I'm a little scared of this because it was just ratified by the IETF in October. So hopefully whichever carrier your IPv6 address lives on runs a good NAT64 cloud, if not your service is going to suck, because the IPv6 network is a rat infested ghetto. Even then, contacting the v4 internet from v6 is going to not work well because there's no traffic on it today. Things like people converting their website to v6 but still using hosted akamai content which is v4, so you can't get the videos, content, etc. The CSS is v4, so the formatting comes up like 1995 web pages. Goodness. Plus websites are going to see the address of your NAT box rather than yours, so maps will think you're in Ohio coming from the time warner location, etc. Personal predictions: -Slow moving Y2k. Companies are going to start getting things like contracts that require IPv6 requirements, they scramble to keep these, throw money and people at it, etc. -IP address cost goes up dramatically. -Lots of people make money on the conversion - people who can make the v4-v6 translation happen smoother (hopefully cisco), consultants, etc. -Timelines - hard to say. I do think the official registries will run out within a year from now. They may open up some previously restricted addresses, etc. Companies start to talk about it this year. Think about it next year. Are forced to do it the year after. This could change so quickly though. Time when we see IPv6 addresses to consumers in the US? Probably at least 2 years out. I think we'll see things like double NAT way before we see IPv6 addresses given to consumers. IPv6 on mobile phones in the US - maybe a year and a half? Asia will lead with IPv6, thank you China/India and their mobiles. -General internet innovation slows slightly. When it comes to implementing v6 it will be a quagmire for many. -Lots of general purpose companies with the premise of - 'you keep your v4 and we'll handle your v6 conversion' pop up and help small/medium and even larger businesses. Congratulations if you're one of the three people that got this far. PM if you want to get into any real details.
IPv6 is coming, albeit slowly, and all those preaching Y2K-like repercussions need to take a moment and remember, well, Y2K. FACT: A large number of content providers have had their content available via IPv6 for a long time.You can go on ebay and buy yourself a little Cisco 871 router and old 2900 switch, and go here and set yourslef up a cool little tunnel and browse the web, IPv6 style. In my opinion, the reason this is so slow moving is because it's so costly. ISP's will need to have dual-stack infrastructure to support BOTH IPv6 and IPV4. In the end though, I don't see end users being affected nearly as much as the doomsday sayers.
As someone who was doing enterprise ops before Y2K, it enrages me every time skeptics talk about Y2K. You know why nothing bad happened on New Years Even Y2K? Because a shitload of geeks busted their fucking asses for months beforehand. Because a shitload of geeks went to their bosses and said 'Look, this isn't me wanting a budget expansion, this isn't a paranoid geek thing, we need to spend money in fixing this, right now, or this business is going to tank when the IT goes under.' and enough of us had that conversation that we actually got shit fixed. Projects I personally worked on prevented a major bank from updating all of their accounts as if a century without action had occured. Without the work we did, every investment deposit would have accrued a century worth of interest and and every mortgage would have showed 100 years worth of late payment fines and interest charges. This was one of the biggest banks in Australia and had billion dollar loss implications. Another project I was peripheral too would have shutdown a major international airport's air traffic control office. Showstopper bug where their comms management system just went offline on the spot and killed every active bit of radio communication on the site. People would have almost certainly died as a result of that one. I worked on a dozen or so other issues closing out Y2k issues - mostly not as serious, but all of them would have had six or seven figure loss implications without our work. Most of them were plausible business killers that would have shut down the business entirely. IPV6 is a less immediate and catastrophic need - but if we fail to address the issue now - there is a real and legitimate implication for business viability in the future. Especially new businesses. There's a reason that people are spending serious money on dealing with this shit now.
You're right, and I should have clarified in my post that I fully understand why Y2K didn't cause the major problems it could have. My point was that IPv6 migration is going to be a slow and smooth transition. Every business from banks to Porn.com are depending on a seamless transition to preserve revenue.
If all goes well - it will look seamless to the end user. There are going to be a lot of business that have this 'oh shit' moment where they realize that neglecting advice and consultation about the future of their network has a cost. Example: a customer that has been working with government and DoD contracts has to turn down a multi-million dollar contract because they can't meet the IPv6 specifications laid out in the agreement. I can't say I've seen a multi-million dollar contract, but I've seen similar to make something like that a reality in the upcoming year or two.