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

[Mark Andrews]

In message <CAO42Z2xo8cDu7_hQC9FoHftMRfWHUttJWvANOh0adV3bRK6Tag at mail.gmail.com>, Mark Smith writes:
> Came across this tweet this morning, which is suggesting that Google
> just purchased a /12 from Merit Network.
> 
> https://twitter.com/atoonk/status/858473247368593408
> 
> 
> The cost per IPv4 address has been in the order of $10 each, so that
> is a purchase in the order of $10M. That's a small amount of money for
> Google, however in real terms, it is a high opportunity cost - $10M
> will buy a lot of other things (staff, servers etc.).
> 
> What has happened is that ISPs that haven't deployed IPv6 have managed
> to externalise the cost of not doing so to Google. Most if not all of
> Google's services are reachable over IPv6, so they're not really doing
> this because they themselves have no alternative way to provide access
> to their services. Google are now receiving ~15% of their traffic over
> IPv6:
> 
> https://www.google.com/intl/en/ipv6/statistics.html

I don't think it actually costs Google that much to be dual stack
over IPv4 only.  The real cost shifting is onto ISP's that don't
have enough IPv4 addresses to supply one to all their customers.
They are the ones being forced to deploy CGNs of one form or another
(NAT44, NAT64, DS-Lite BR, ....).

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.

> Others who want to continue to grow and continue to provide their
> services to IPv4 only clients will also have no choice but to buy more
> IPv4 addresses. Other larger players will be able to afford to do so,
> smaller players may not be able to.
> 
> At some point in the future I think there will be push back from
> organisations who are being forced to or can't afford to buy IPv4
> addresses because ISPs haven't deployed IPv6. They'll want to somehow
> shift some or ideally all of the costs or consequences back onto the
> organisations who are forcing them to buy more IPv4 addresses.
> 
> I think they're unlikely to try the service charge approach, although
> from the side line, that might be both interesting and entertaining to
> see if it works. I'd like to be there when an ISP receives the first
> "IPv4 service charge" bill from an Internet service giant to see the
> look on their face. Alternatively, per-customer service charge could
> be used when the customer is paying for the service - e.g., it could
> be a line item on the Netflix bill.
> 
> I think the only other lever these IPv4 purchasers have is to degrade
> the IPv4 service - either artificially, or by choosing to avoid
> capacity/resource investment in it, as it is a legacy service. ISP
> customers call their ISP's helpdesk when ever anything goes wrong, and
> that is where the cost of not deploying IPv6 will be incurred.
> _______________________________________________
> AusNOG mailing list
> AusNOG at lists.ausnog.net
> http://lists.ausnog.net/mailman/listinfo/ausnog
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: marka at isc.org