[AusNOG] Prediction: Google et. al. may artificially penalise IPv4 clients

Mark Smith markzzzsmith at gmail.com
Mon May 1 12:32:37 EST 2017


On 1 May 2017 at 10:31, Mark Delany <g2x at juliet.emu.st> wrote:
>> That said rolling IPv4 blackouts by Google, Netflix, etc., would
>> attact attention of the press and ISP's that aren't dual stacked.
>>
>> Whether they would be willing to do it would be another matter.
>
> Exactly. The PR risk is huge. Regardless of their altruism, US-based
> 800lb gorillas imposing their will on under-resourced non-US ISPs is
> not a message you want sprayed all over the twitter-verse.
>

These under-resourced ISPs may already be zombies.

IPv6 has been coming for a very long time, so they can't really claim
to be unaware of it. If they haven't got enough funds to cover their
IPv6 deployment, they're probably already in financial trouble but
just don't know it yet.

Even if they say they can't afford to deploy IPv6, they still won't be
able to avoid the cost of either buying IPv4 addresses or buying CGN
capacity to continue their business (and buying CGN capacity is
unavailable if you run out of IPv4 addresses, even if you deploy IPv6.
It's the amount you have to buy that changes if you deploy IPv6 to
customers). Either way, additional money will need to be spent over
and above normal and current running costs.

> It's a devilishly difficult problem because it's hard to reward v6
> adoption without punishing v4 recalcitrants. And who wants a
> reputation as a punisher?
>

Punishing v4 recalcitrants is fundamentally unavoidable. Their
continued existence would not be able to subsidised endlessly by the
rest of the industry continuing to buy IPv4 addresses, effectively on
their behalf.

I'm not saying there will be rolling IPv4 blackouts. What I am
suggesting is that IPv4 clients may purposely receive a degraded
service compared to those using IPv6.

The way to motivate action is consequences. The consequences can be
good or bad ("the carrot or the stick"). Better IPv6 performance/worse
IPv4 performance will be be a carrot for IPv6 and a stick for IPv4.

The reason why I think Google buying the /12 is significant is that
Google are almost if not entirely thoroughly IPv6 enabled and have
been for quite a while. So they're not buying those IPv4 addresses to
solve their own lack of IPv6 deployment problem, they're having to do
it to overcome others lack of IPv6 deployment. They're paying a large
amount of money to mostly solve somebody else's problem rather than
their own. I can only see them and others in a similar situation
tolerating those costs for a limited time. They have a financial
motivation to actively minimise or avoid those costs sooner rather
than later.

Regards,
Mark.


More information about the AusNOG mailing list