Jeff Pulver proposed an interesting idea called IPv6 Day:

Collectivity we should pick a day, at some point in the near and
foreseeable future and push everyone to reboot themselves and make it such
that from that point forward, IPv6 will be supported on all networks which
interact with the public Internet.

In geeks term, we call this a ‘flag day’. The last time we have a flag day was 1st Jan 1983 when Internet moved from NCP (Network Control Protocol) to IPv4.

So why not do it for IPv6?

Well, first of all, there are far far less hosts connected to the Internet back in 1983. Today, we have all kind of IP devices, some of them requires firmware update or total replacement to go to IPv6. A flag day for IPv6 would potentially cost as much, if not more, then Y2K. (yes, please laugh)Even if we figure out how to pay for the upgrades, most of the IPv6 routers today are software based except some high-end routers. This means the performance is much less with the IPv6 switch on. This has been one of the major stumbling block in getting ISPs to deploy IPv6. No ISP I talk to is willing to put software routers onto their operational network.

Even with the routers solved (things are looking good in the last couple of months actually), the planning for IPv6 flag day is going to take years. Japan, for instance, has declare that all their network will be IPv6 ready by 2005. All indication so far is they will likely to push back the deadline. A flag day in 2005 is far too aggressive…2010 or 2015 may be more reasonable.

But the most important reason of all, there isn’t a real demand for IPv6 today. While there is significant reasons to move from NCP to IPv4 back in 1983, such reasons does not exists for IPv6. No demand, means no business, means no ISPs will migrate. Ideological appeals like End-to-End are nice but not sufficient to move ISPs who looks at their bottomline.

What about IP address exhaustion? Sorry, that’s a myth. I did a presentation in WTSA in Brazil using Geoff Huston data which projects we don’t run out of IPv4 address until 2020++.

But we do need IPv6. We need to bring back End-to-End so machines can talk to each another Peer-to-Peer, without Man-in-Middle. We need to get NAT out of way so protocol like SIP can be more seamlessly deployed without STUN and other NAT traversing tricks.

As Mao Zedong once said: “Revolution starts from the country side”. Likewise, IPv6 revolution will comes from the edge. More applications should start using IPv6 like Three Degrees. Better yet, have a killer IPv6 SIP application. These will give home users reasons to install IPv6 and use IPv6 on their home network. Create the demand from the users, the rest will follows.

Partial Disclosual: We support one project to create IPv6 SIP. More details later.

ps: This entry has been syndicated on CircleID.