[arin-ppml] The non-deployment of IPv6
On Dec 9, 2009, at 7:16 AM, Chris Engel wrote:
> Well, speaking for myself and the Enterprise I manage..... I see absolutely no reason to do IPv6 at the present time. There is no benefit to it and a fairly decent cost. Furthermore the paradigm is sufficiently different that it's not like it's a seemless transition from v4. If they had just taken 4 and tacked on a few more octets but made no other changes as to how it worked, that might be a bit of a different story. No surprise that many Enterprises aren't rusing to adopt. The key question for an Enterprise Admin is do you have control over the hardware and networks on which the services you provide depend? If the answer to that is ....yes.... then looking to switch to v6 really makes no sense.
Yep... In general, the enterprise desktop will be the last group of users to migrate to dual-stack.
Probably not doing so until forced to by IPv6-only desirable content, or, by running out of
IPv4 addressing resources within the enterprise.
That's OK. The big push right now really needs to be ISP backbones (quickest conversion,
really, and the simplest at this point) and Content providers (also relatively easy).
If those two groups get to dual-stack before runout, then, most of the rest will be a non-issue
as they can migrate at their leisure without much systemic impact.
> The point at which you need to do something about v6 is when you are going to start getting external traffic (SMTP, HTTP, etc.) from v6 only users... or when your own users are going to start to want to access v6 only sites. That, I imagine is still a fair ways off.
It's probably about 3 years away at this point, realistically. I wouldn't count on it being any
further off, and, it could be as close as 2 years, but, I think that's unlikely.
> When that happens what the Enterprise Admin is most likely to look at doing is layering in support for those functions in a way that requires the MINIMUM neccesary changes to thier existing infrastructure. I imagine that will be some variety of v4 to v6 gateway services or something like that. A few years down the road, if some-one is building a new network from the ground up I could see them maybe deciding to go v6 native. However, for existing networks...as long as there are robust v4 to v6 solutions available....and I imagine there will be due to demand.... I can't see any reason to switch away from v4 native for the forseeable future.
There's nothing wrong with this approach. I think that in general, the IPv6 push has done a
poor job of focusing on the critical early points of conversion, and, many people still seem
to have the misperception that this is a migration to IPv6 rather than the addition of IPv6
capabilities to the IPv4 network.