Yesterday's Windows update causes IPv4 to be default

MarcoH - lists mch-v6ops at
Sat Nov 24 13:06:11 CET 2012

> What we're going to see over the next decade are an ever-increasing number of v6 rollouts, many (most?) of which will be repeating the same mistakes over and over again as they come to grips with this new technology. So for user networks that have already implemented it successfully there will be a constantly changing list of content sites that have only recently implemented v6, and have done it wrong. And of course, there will be ISPs with new v6 rollouts that haven't got all the kinks worked out yet, which will affect users of the content sites that have successful v6 implementations today.
> So unless we solve what will unquestionably be the _ongoing_ problem of faulty v6 connectivity (wherever it is located) causing problems for users, in real time, the answer to the problem is going to continue to be "turn off that eye pee six crap."

Hi Doug,

Permit me to cherry pick in your response :) Yes you are absolutely right, their is a fair chance that people will keep on making the same mistakes and every new deployment suffers from teething problems in the first few months/years. Not that dissimilar  from IPv4, MPLS or higher layer problems such as spam, data leaks, phishing or got knows what goes wrong these days.

I've made my set of mistakes in the past and keep on making them, hell maybe this is another one of them. You can't entirely prevent it. It is how you handle the outcome what really matters. Somewhere in my earlier years I ran into a CTO who explained it like "Faults will happen, just make sure whatever caused it is always a unique incident". Take away the cause instead of just fixing the (visible) problem, clearly he was an engineer himself.

This is easier said then done unfortunately. You can of course educate people, but that takes time and money. And of course there will always be a subset of people who don't want to invest or are convinced they can do it without any training. I admire such dare devils and maybe they get lucky (I once was :). After all you need the pioneer that takes a bite of that new fruit somebody found, with the risk instantly dropping dead. As long as you give people some feedback on the outcome it serves a purpose.

On the point of investing in IPv6, training is not the only element that suffers. And as engineers we can not really solve it. What we can do is try and educate the people who have to sign off on the money. And to proof we are right it helps if there are a certain number of people that can be considered pioneers and are willing to take the risk and invest in unproven technology.

I have the feeling a lot of the people on this list are in that fortunate position that they were allowed to experiment and run some risks with this all brand new protocol. Some of us were even lucky enough to do this in a time when the Internet was less of a utility service it is today. We could break things for a few hours without a number of banks going bust and causing a global economic recession.

We've been messing with this for over 20 years and learned some valuable lessons. But life in a petri dish in the protective environment of a lab is different from the world outside. There are no predators around such as profit margins or time. So reality is that things that might have worked in a lab, fails when you push it out to a million subscribers. Which means you have to get back and try again, but right now time has made it way into your lab and is willing to eat your experiment even before it reached adulthood.

What we need is for those engineers who have experience to stand up and report back. Of course there are multiple reasons why you shouldn't. As you just spent 20 years in lab you may have some competitive advantage. Maybe you even think you made it work and are now making some money back in the form of license or upgrade fees. The last thing you want is a bunch of customers who want their money back because it turned out you had it wrong. And even a worse outcome would be for your competition to learn from your mistakes and come back with a cheaper and better alternative.

If we think the transition to IPv6 is a collective problem we need to collectively sort it out. We need people to swallow their pride and tell the world what went wrong. And importantly tell people what it is really like, showing the scars and make up some battle story instead of admitting you shot yourself in the foot is counter productive. Not everybody can be a hero and poor blokes who step out there thinking they are invincible, are probably the first one fatally wounded.

We need to educate people, but more importantly we need input on what to educate them about. Sharing experiences is what will ultimately will solve all these issues. Make sure that the next guy won't make the same mistake you did. If your transitioning technique turns out not to be the best, supersede your RFC with your experiences and move it to historic. Have manufacturers focus on what probably works instead of trying to implement everything at once. That way they might even have time left to actually fix a few bugs. Share your experiences and admit you got it wrong. The people who do, are the real heroes.

This is the real problem, there aren't that many people around who are willing to do this. Wether it is matter of ego, fear of competition or a manager who says no, because he thinks it is a waste of money. As long as we refuse and try to keep our knowledge to ourselves, others are forced to invent their own wheel and ultimately break things. Which in turn cause others to setup some defensive mechanisms, like the one which started this or even worse and don't do anything at all. Stick to what you know and "proven" technology like NAT.

We are at the crossroad and have to make a choice: Implement IPv6 globally and keep the net open and free or keep driving people towards IPv4 and NAT and in the end find ourselves locked behind some paywall. We don't need the ITU to take over, we are perfectly capable of breaking the Internet ourselves.



"Good tests kill flawed theories; we remain alive to guess again"

More information about the ipv6-ops mailing list