Re: [v6ops] Time to dump dual-stack networks and get on the IPv6 train - with LW4o6

[George Michaelson <[e-mail]>]

I don't want to over-play it, but some of this goes to "the death of
transit" -IPv6 works fine as a direct relationship, or to content in
the CDN, especially if the CDN is tight to your ISP.

But there are downsides, like the loss of end-to-end in this. its a
model driving to vertical capture, where the CDN players and your ISP
own you and nothing happens normally beyond those boundaries.

I *like* immediacy,  I'm no different to anyone else @home, I like
short RTT and jumbogram service delivery. But, lets be careful we
don't forget packets have to go outside, and when they do, IPv6 still
has some issues in transit: There are far too many tunnels, and far
too many routes where it bounces, trombones, or just doesn't work.

-George

On Mon, Jul 2, 2018 at 4:28 PM, Fred Baker <[e-mail]> wrote:
> On Jul 1, 2018, at 7:53 PM, George Michaelson <[e-mail]> wrote:
>> I know its "wrong" but I sympathise with a tech journalist seeking
>> clarity in this. The edge user situation is very hard to distinguish
>> in both cases, the user functionally can see both Ipv4 and Ipv6
>> resources. How is this not "dual" in the sense a man or woman in the
>> street would understand it?
>
> Of course. From the perspective of a residential - or enterprise - user, they have both IPv4 and IPv6, and the network is carrying them. So from their perspective they retain dual stack capabilities in the near term. However, here's another factoid.
>
> For the residential user, the number of services that actually require IPv4 is relatively small. I mentioned home security as a market that required the ability to make an IPv4 connection to the home. Suppose that could be changed? One way to change that would be for a few large ISPs to convince companies like LINK or ADT to become IPv6-capable, whereupon it becomes a market requirement for their competitors.
>
> The ISP has just made having IPv4 at all a second priority for them; it's in their interest to move residential/soho/enterprise to IPv6, with the goal of removing the final translation step. I would expect their next move to be to charge for the service, and if possible move it to the edge itself  - which is where draft-*-v6ops-nat64-deployment is going. The transition rolls out from there.
>
> The more outfits that find it in their interest to be able to operate without IPv4 at all, the better this is for all of us. And this becomes an important step in the general transition.