[ppml] getting converts to V6

[Iljitsch van Beijnum]

On 21-mei-2007, at 5:48, Lee Dilkie wrote:

> The ISP cannot simply force end
> users to take IPv6, there are hardly any working applications that  
> work
> on that network at present.

That's not true. Some examples of applications that work with IPv6:

- Internet Explorer
- Windows Media Player
- Firefox
- Thunderbird
- Quicktime Player
- iTunes
- Safari
- Apple's Mail
- iChat (limited)

> It's not about "providing the service", you
> need to create demand first. (Can you imagine starting up a TV  
> broadcast
> station before the TV itself was invented?)

It's more like the addition of color, technically speaking. For that  
to work, viewers must have color TVs, stations must broadcast a color  
signal and content must be produced in color. You need all three,  
just one or two won't give you any benefits. Now of course the  
difference is that color is highly visible to the user, while IPv6  
isn't. How can users be expected to go ask for something that they  
don't even know exists? Waiting for user demand is pointless here.

> And all this fits how, you ask? The original email talked of trying to
> get folks converted to IPv6. I pointed out that, contrary to some  
> other
> opinions, it's not the ISPs that need converting (though they  
> certainly
> do, but they will anyway to follow the money) but rather the industry
> (the "internet") needs to create demand to move to IPv6 with some
> as-yet-discovered killer IPv6 application that will make the switch a
> compelling one for the end users.

Ah, the mythical killer app. There are basically two types of  
applications, and IPv6 can't be a requirement for either of those:

- client/server apps: these work without problems today, nothing IPv6  
could add
- peer-to-peer apps: don't work so well today, but requiring IPv6  
means the app is dead in the water

Or in other words: in today's world, it's always more beneficial to  
get something working over IPv4, even though that may be hard and  
unreliable, rather than to get the entire user population to adopt IPv6.

To get IPv6 adopted you need to have IPv6 capability in the following  
places:

1. user side application
2. user OS
3. SOHO gateway/CPE
4. ISP networks
5. content side firewalls and load balancers
6. content side OS
7. server application

By now, 2 and 6 are not a problem. 1 and 7 aren't always there yet,  
but there is enough software that supports IPv6 to make a good start.  
4 is largely doable, although there is probably still some equipment  
out there that can't do IPv6 at all or not at the same speeds as  
IPv4. But the real problems are 3 and 5.

I have native IPv6 connectivity over ADSL. However, this required me  
to pay several times more for my ADSL modem in the form of a Cisco  
826/827 and it required manual configuration on both ends of the ADSL  
link. With IPv4, you can pretty much buy any ADSL or cable modem,  
plug it in and it works. Not so with IPv6, because there is no  
standard way for ISPs to provision the CPE boxes with an IPv6 address  
range. This is very similar to the situation we had when ISDN first  
came available in the mid-1990s: everyone used different protocols so  
it was very hard to get it working. But after a while a standard way  
to do it came about and ISDN became easier to deploy than analog modems.

I don't have a very good view of what's going on on the content side,  
but as far as I know, the fancy stuff used to serve up content to  
huge audiences can't do IPv6 on the same level as IPv4 right now.

So what we need to do:

Software makers: make your software IPv6-compatible.

Content people: push your vendors.

Cable and DSL ISPs: talk to the CPE vendors and work out a standard  
provisioning model. (Hint: DHCPv6 prefix delegation can work wonders  
here.)

ISPs in general: for now, set up a tunnel broker for your customers.  
That gives those who care enough to manually configure it most of the  
benefits of IPv6 but it's only a fraction of the hassle. Also set up  
a private 6to4 relay to provide a better experience for customers who  
use automatic 6to4 tunneling.

"Ask not what IPv6 can do for you; ask what you can do for IPv6."