On 12 dec 2007, at 17:24, Joe Touch wrote:
Lars and I talked originally about a 'doomsday clock'-like metric,
that
shows IPv6 uptake via tracking Fortune-500 and similar commercial
sites.
This is the current result... not very encouraging, IMO.
You leave out what the result is, but I assume it's only a handful, if
that.
Which is neither surprising nor a problem. There are three issues with
making a big website IPv6-enabled:
1. Many of those use complex load balancing mechanisms to provide one
service from a number of physical servers. It seems the vendors of
these products aren't very quick with their IPv6 uptake and even if
that isn't a problem, it's non-trivial to replicate these mechanisms
for IPv6.
2. A small number of systems that think they have IPv6 connectivity
don't actually have IPv6 connectivity. The most prominent example of
this is Windows Vista machines that automatically enable 6to4 when
they have a public IPv4 address but some networks that do provide
public IPv4 addresses filter IPv6-in-IPv4 packets.
3. Although this has been getting a lot better the past years, it
still happens that IPv6 connectivity is slower or buggier than IPv4
connectivity.
As long as all the eyeballs are IPv4-capable thhere is almost no
incentive for the content people to invite the problems above, even if
the service provider hosting the content does support IPv6. (For
instance, check to see how many people peering at the Amsterdam
Internet Exchange have an IPv6 address for their router, and then
check how many have an IPv6-enabled website.)
_______________________________________________
Int-area mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/int-area
