On killing IPv6 transition mechanisms

[Ted Mittelstaedt]

Ole Troan wrote:
> Ted,
> 
>>>> I wonder, how no connectivity could be better than some?
>>> because content providers measure the difference between IPv4 and
>>>  IPv6 behaviour for dual stack hosts. if the number of hosts with
>>>  broken IPv6 connectivity or with significantly worse latency
>>> than IPv4, they will not enable IPv6. to put it bluntly: if you
>>> don't get IPv6 from your SP, then don't bother. you are doing
>>> more harm to IPv6 deployment than good.
>> Please quit FUDDing.  Content providers only care that customers on
>>  the Internet are able to get at their offerings.  Ebay, Craigslist
>> and Amazon don't give a crap about latency of a connection if they
>> get a credit card number over the connection for a valid order.
>> And there are still a VERY LARGE minority of people on dialup-only
>> IPv4, so quit being a connectivity snob - the worst IPv6 broadband
>> connection out there is going to have lower latency than a 56k
>> dialup modem!!!
> 
> see:
> https://sites.google.com/site/ipv6implementors/conference2009/agenda/10_Lees_Google_IPv6_User_Measurement.pdf?attredirects=0
> 
> 
> which shows that a dual stack host is about 150ms slower using IPv6
> than IPv4, and that ~0.08% of the hosts have broken IPv6. (please
> take notice in the presentation that there is quite a bit of
> statistical uncertainty because of few IPv6 clients).
> 
>> While the IPv6 deployment isn't served by "broken IPv6
>> connectivity" any content provider looking at IPv6 knows 2 things,
>> first that most if not all IPv6 connections to their dual-stacked
>> servers are "beta testing", (since IPv4 is still available from the
>> RIR's)  second that the more connectivity options they have to
>> their servers the greater the chance that someone will be able to
>> connect to them and get their content.
>> 
>> You seem to assume the content providers on the Internet are
>> idiots. They know that IPv6 right now is in it's infancy, and they
>> are watching the usage of it.  What is more important right now is
>> GETTING the IPv6 connections from users, not how good they are.
>> That's what they are paying attention to.
> 
> if I was a content provider I would look hard at those numbers above
> to judge if I wanted to piss off 0.08% of my customers and slow down
> the web site for a quarter of a percent of the others.
> 

Your thinking in terms of a garage operation ISP.  "dual stack" to a
content provider means "dual stack the SITE not the HOST"

Most if not all the large content providers have multiple servers
behind a load balancer.

It's pretty obvious that devoting merely 1 of the servers out of their
server pool to IPv6-only is the way to go, here.  In fact you would
just put it on the Internet directly, not behind the load-balancer.
Once IPv6 traffic is large enough you would create a separate pool of
IPv6 servers behind an IPv6 load-balancer.  Dual-stacking the host is 
not needed to dual stack the site.

Of course, I should not assume to tell content providers how to
run their businesses, if they want to do it the bass-ackwards way
that's their right. :-)

> if we continue to push for mechanisms which have higher probability
> of delivering broken IPv6 connectivity and/or high latency then I'm
> afraid that content providers will be reluctant to enable IPv6.
> 
> and users will disable it. see: 
> http://www.google.com/trends?q=disable+ipv6%2C+enable+ipv6
> 
> (and no, not even I pretend that the above is scientific, but users
> shouldn't search for "disable ipv6" at all. ;-))
> 

Only the "power" users are going to bother digging that out and only
in response to problems they perceive are happening with their
machines.  And these are the people who are probably going to be
among the first and last groups on IPv6.  And as long as their ISP's
are running current nameserver code, they shouldn't be having
problems, thus they will make the registry change, and nothing
will happen, then 3 months later when they scotch their machine
due to excessive tinkering with it and
reinstall windows, they will not bother making the change

Unfortunately this is all part of educating the consumer about
what constitutes a decent ISP service and what doesn't.

The fact of the matter though is that it really only matters
to get the center of the bell-curve users on IPv6.  Once that
happens the power users will have to get on it also.

Ted

> cheers, Ole
> 
> 
>> Lastly, SP's aren't going to offer IPv6 unless they see monetary
>> loss by not offering it.  Obviously the best way is customer loss
>> but there are others. Currently, Ivan's VPN solution essentially
>> REWARDS his current non-IPv6-offering SP - however, Ivan can easily
>> counteract that by calling into his SP, weekly if necessary,
>> demanding IPv6.  It costs a SP money to pay labor to answer the
>> telephone, and if Ivan makes 2 phone calls a month and gets a live
>> human being at his SP on the phone for at least 10 minutes,
>> "filibustering" for IPv6, (ie: talking on and on an on touting his
>> IPv6 VPN solution and how much better native IPv6 would be) then he
>> will have eaten up all profit for the month for his SP on his
>> account - and if enough people do this, then even SP's who are
>> hostile to IPv6 will be forced to deploy it.
>> 
>> Ted
> 
>