On Wed, 31 Mar 2010, Nicholas Weaver wrote:

:: 
:: On Mar 31, 2010, at 12:28 AM, Igor Gashinsky wrote:
:: > 
:: > You are absolutely right -- it's not a DNS problem, it *is* a host 
:: > behavior problem. The issue is that it takes *years* to fix a host 
:: > behavior problem, and we need to engineer and deploy a fix much sooner 
:: > then that (hopefully about a year before the v4 exhaustion date). Given 
:: > that, is there something other then DNS that can address it better/faster?
:: 
:: Except that, since it IS a host problem, why not focus on improved detection 
and debugging....
:: 
:: EG, yahoo is one of the big pushes behind it.  THey could easily add 
auto-debugging to their homepgae would would detect this condition and NOTIFY 
the users that this will be a problem, same for google.

Yes, we can identify the users (this is how you get the stats in the first 
place), but afterwards, the problem then still becomes 2-fold. There is 
a *very large* subset of users who will ignore any warnings (see attempts 
to deprecate IE6), and there is another subset of users who will do just 
about anything that the the websites asks them to do (those who get 
phished all the time), and then there are a ton of people in-between. 

So, assuming the user doesn't ignore the warning, and is prompted to 
action, what do we tell the user to do? We don't know what's breaking v6 
on their system/network, and while we can put together an FAQ that the 
more technically inclined may follow, the majority of users won't know what 
it's talking about, so, do we tell them to call their ISPs, or contact 
GeekSquad or somebody and pay them to fix it? 

>From an ISP perspective, that may not be what they would want to do 
(support calls cost a lot of money), and from a user perspective, they are 
going to go "what is this ipv6 thing that's breaking, why do i care, and 
why should I waste money trying to fix it", or worse, "why the hell is 
$content_site probing my pc, I'm going to contact the FEDS!" (you'd be 
surprised how often that happens)..

This is why we are proposing this hack -- it will allow an ISP to turn on 
ipv6 on their network without worrying about breaking existing IPv4 users 
(with the caveat that it will prevent some v6-capable users for using the 
ipv6 internet until they can v6-transport DNS queries, and the negative 
impact on DNSSEC). And, it is going to be that operator's choice if they 
want to enable this feature or not.. It has also been pointed out to me by 
somebody who wishes to remain anonymous that I have not adequately 
described the benefits for deployment of this feature to ISP's, so quoting 
his message:

"Let's say that wireless Service Provider A has done their home work and
implemented Ipv6. Let's say that Wireless Service Provider B is happy
with IPv4 and have no plans to implement IPv6. There are all sorts of
devices, operating systems and versions in the network (Symbian,
Android, Linux, Windows, MacOS, PalmOS, embedded devices, aircards,
etc). Most of them are not field upgradable (or very hard to upgrade).

A user will not be able to tell why it takes 5 seconds longer to get to
Facebook on Service Provider A's network compared to Legacy Network B.
The user will blame the Network and probably switch to Legacy Network B.

A solution like the one you are proposing will allow a service provider
to roll out Ipv6 without fearing increased customer churn."

Fundamentally, what we are proposing is a way to ease the transitional 
pain, and making it an optional feature that the operator may enable, if 
they so chose (and see that the benefits outweigh the drawbacks). Is there 
a better way to address the problem?

Thanks,
-igor

_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to