Re: Just Thinking (About the Nightmare Transition Ahead)

[Roman.Arcea@xxxxxxxxx]

Hi,

I agree with Brian. What nightmare? It only seems as a nightmare until you don't have it in place.

It is not about the transition itself, rather about the conceptual fact. First off all, the common belief is for 0.05% of brokenness (most of it coming from old version of Mac OS X, some from old versions of Opera etc.). Now, that was half a year ago. Is it the same today? Have you noticed that no one has ever said that the service WILL be broken, but rather MIGHT be broken. This is a very important distinction. Also, do you know of any standardized method of measuring it? Have you ever measured the brokenness in IPv4 due to MTU issues or DNS problems? For instance I have quite often encountered issues with my home service provider, that had an overloaded DNS during the peak hours that used to time out on various occasions, and the page would not display until you hit refresh. Is it brokenness?

So, really - I believe that to some extent the World IPv6 Day is a first step that proves QA departments are starting being comfortable with the new realities.

What is a little hard, is the operations, where you need to watch now for two protocols in the same time, while the feature parity is still an issue. But I would hardly call this a nightmare.

Best,
Roman


-----ietf-bounces@xxxxxxxx wrote: -----

To: Brian E Carpenter <brian.e.carpenter@xxxxxxxxx>
From: Sabahattin Gucukoglu
Sent by: ietf-bounces@xxxxxxxx
Date: 01/22/2011 10:47PM
Cc: ietf@xxxxxxxx
Subject: Re: Just Thinking (About the Nightmare Transition Ahead)

On 22 Jan 2011, at 18:48, Brian E Carpenter wrote:
> What nightmare? I find IPv6 dual stack works just fine.

It does, when your connectivity is working.  Unfortunately, the 0.1% (or whatever it is) of users whose connectivity isn't working seem to be sufficient in number to prevent large sites from deploying, and a number of others to just turn IPv6 off.  Not news, except I'd hate for this to be justification in keeping dual-stack from happening (the effects of which must be considerably worse than just losing 0.1% of users), especially since the publicity about "World IPv6 Day" has brought the whole issue to light.

Of course, I'd also like Google to run IPv6 (on its main web servers) for longer than 24 hours at a time. :-)

> However, see draft-wing-v6ops-happy-eyeballs-ipv6

Yes.  I'd love there to be some sort of consensus on it, or something like it.  I'd focus more on testing connectivity, though, rather than optimal calculation of IPv6 reachability using one protocol/application to any given place at the time (i.e. rough and ready determination at OS boot whether we're on a broken network, EG).  It would not be foolproof, but it would not need to be in order to be useful in the short-term.  Perhaps Happy Eyeballs is the answer.

Cheers,
Sabahattin
_______________________________________________
Ietf mailing list
Ietf@xxxxxxxx
https://www.ietf.org/mailman/listinfo/ietf

_______________________________________________
Ietf mailing list
Ietf@xxxxxxxx
https://www.ietf.org/mailman/listinfo/ietf