13 April 2011 by Ch. Tronche | Comments Off | Filed in Divers
This is part of my series on (Trying to) foresee IPv6 deployment.
Google is doing a lot of things for IPv6 adoption with its IPv6 program. First, it brings credibility to IPv6. When Google talks, the world listens. They have hosted a workshop about IPv6 adoption, where at least two important points where made:
about 0.5 % of internet users have a broken IPv6 connectivity, that is their computer thinks it has IPv6 connectivity, if presented with an IPv6 address, the computer will try to use it, but it will ultimately fail. Depending on your view, 0.5% may be acceptable or not (not acceptable from the point of view of Google)
T-mobile USA brought steam to the NAT64 approach by saying this was their way, and by putting some hard figures about NAT64 brokenness on the table.
However, as an ISP, for your clients to use Google (and affiliated sites like Youtube) over IPv6 by default, you need to have a special agreement with Google. This is because of the 0.5% that Google doesn’t want to lose. The point is that, for now, most ISPs still have a stock of IPv4 addresses, so their new customers have a dedicated IPv4 address to get the best possible connectivity. For these ISPs, there’s absolutely no incentive to enter into the agreement with Google. On the other hand, if IPv6 access was enabled by default, it would move large amounts of traffic to IPv6 for the ISP customers that have IPv6 enabled, pushing the ISPs towards IPv6.
All in all, I think the current Google policy of “IPv6 on agreement” restrains rather than encourages IPv6 adoption.
2 May 2010 by Ch. Tronche | Comments Off | Filed in Divers
… But the adventure goes on on the next generation, Tronche 3.0.
26 January 2010 by Ch. Tronche | 1 Comment | Filed in French politics, techtalk
Tracking of internet users on the basis of their IP address seems to have some traction theses days (cf. for exemple the HADOPI law), at least in France.
With the coming exhaustion of IPv4 adresses, Internet Service Providers are putting together “transition” (permanent ?) mechanisms to make IPv6 works (when such a service do exist), with as little disturbance as possible for the existing users.
Part of these mechanisms may be to NAT IPv6 addresses distributed to their users to IPv4 adresses to keep the existing IPv4 services working. In order to do this, the ISPs will have to get the already distributed addresses back to feed the NAT mechanism.
The result: an IPv4 address will be used in succession by various users at different moments. The IP address (v4) alone won’t be enough to identify the user any more, only a set of dozens of clients.
Of course, there is a solution: to identify the user, the source port will have to be noted together with the IP address. This implies the ISP will store this information, a heavy and costly process, since the NAT associations are changed much more often than the IP addresses.
Simple technical difficulty or yet another risk of error ?
Timing can also be called into question: putting an authority like HADOPI in motion can take years, while the IP address exhaustion is predicted around 2012, forcing ISP into quick action.