I motivi del ritardo nel passaggio da IPV4 a IPV6

Ha ragione Gabriel Ramuglia ad elencare i motivi del ritardo nel passaggio da IPV4 a IPV6 e fa bene a denunciare che probabilmente c’è tutto l’interesse degli operatori a spalmare la procedura su 10 anni. Una sola cosa l’articolo non dice. Che i veri responsabili non sono gli ISP ma chi potrebbe governare questa transizione e non lo fa. Il Governo, quindi, con atti di indirizzo politico, potrebbe  accelerare eccome l’adozione del IPv6. Ma intanto non si fa e si continua a nattare. Niente di nuovo, sia chiaro. Io stesso ho partecipato a decine di riunioni a Viale America, Dicastero Comunicazioni. Ma niente è cambiato.


Why IPv4 is Here to Stay

IPv4 depletion continues to be a hot topic in technology. When and how the transition to IPv6 will occur is something of great concern to the hosting industry in particular. In part one of this series, I made the case that IPv4 will continue to be mandatory for hosting providers until all of the large residential ISPs enable IPv6 for 100 percent of their customers.

This begs the question: how quickly will the large ISPs reach 100-percent IPv6 support for their customers? In part two of this three part series, I make the argument that ISPs have a strong financial incentive to delay the IPv6 transition as long as possible.

There are some well-known problems in transitioning to IPv6: it requires you to replace hardware, re-train employees, re-write software, etc. We won’t cover those issues today because I expect these problems can be solved within five to ten years.

Instead I’d like to discuss whether ISPs stand to make money by delaying IPv6 adoption. I believe ISPs can substantially boost their bottom lines if IPv6 adoption is delayed.

In order to understand the financial incentives, there are three relevant questions:

How many IPs do ISPs already have?
How many IPs do ISPs actually need?
How much money are these IPs worth?
First, how many IPs do the big ISPs already have?

Looking at six of the largest residential ISPs (Comcast, ATT, Verizon, Cox, Time Warner Cable, and Century Link), you can see that they are originating at least 238 million IP addresses from their respective AS’s (AS7922, AS7018, AS701, AS22773, AS11351, AS11426, AS11427, AS7843, and AS209). It is worth keeping in mind that the US only has 115 million households in it, giving these ISPs alone two IP addresses for every household in the US.

Comcast is originating 71 million IP addresses from AS7922. To put this in perspective, Comcast has only 53 million active services:

2013 Comcast Subscribers:

21.7 million video customers

20.7 million high speed internet customers

10.7 million voice customers

Comcast is using more than one IP address per service subscription. With many households subscribing to more than one service, there are several IPs in use for every household.

The reason these ISPs use so many IPs is simple: ARIN will give ISPs like Comcast more IP addresses so long as they can show that they are using what they already have. This gives ISPs an incentive not to conserve IP space and to hoard as much as they can before ARIN runs out. This lets them stockpile IPs today and hold back any “IP efficiency projects” for later.

This leads to our second question: how many IPs do ISPs need in order to run their networks?

Nobody knows exactly, but a lot fewer than they have now. Though there are disadvantages to NAT, consumer ISPs could enable NAT for most customers. NAT is already used every day in nearly every home and business, and works well for the majority of typical Internet subscribers.

If you’re skeptical that NAT will work well enough, I can’t say that Comcast agrees with you. In their 335 page annual report, the terms “IPv4” “IPv6” and “IP address” are contained in the document zero times. However, it does list many mundane and unlikely risks such as: “We may be unable to obtain necessary hardware, software, and operational support” (from third-party vendors). Comcast is not stupid, and realizes something many don’t: IPv4 depletion poses no significant risk to their business, and may even strengthen it.

Why is this? With a NAT ratio of just one IP per four households, all US residential internet use would require under 30 million IPs, 1/8th of what the big six ISPs control. With Comcast alone having 71 million IPs for 21 million internet subscribers, a 4 to 1 NAT ratio would imply Comcast has 65 million excess IPs. To understand why one or more of the big six ISPs may drag their feet on the IPv6 transition, we need to understand the value of these IPs.

This leads to our final question: how much money are these IPs worth?

Today, hosting providers generally charge end users around $1/month for extra IP addresses. A small host will pay a similar fee to their data center for these IPs, whereas larger hosts get these IPs for almost nothing from ARIN. If larger hosting providers had to pay for IPs instead of getting them from ARIN, Comcast’s IP hoard alone could already fetch over $700 million a year in lease fees. This would already be nice boost for Comcast, whose video, voice and internet divisions together had $10.8 billion in profit in 2013.

More interesting is when ARIN stops giving out cheap IPs. In part one I argued that a hosting provider would still support IPv4 even when paying $100/month for each IP. If the large ISPs drag their feet on an IPv6 transition, the fair market value of IPs could easily reach $10/ month, and perhaps even $100/month.

At $10/month per IP address, Comcast’s excess 65 million IPs could be worth $7 billion a year to companies like Microsoft, Amazon, or Digital Ocean, who have a relentless need for more IPs to grow their cloud services. This would nearly double their annual profit without requiring any additional investments.

Nobody can predict the future, and even less so can you predict the actions of top management of very large companies. Nevertheless, for an ISP in this position, the temptation will be very strong to drag out the IPv6 transition for as long as possible. If even one of the six very large US ISPs refuses to fully support IPv6, then IPv4 will remain in place for far longer than anyone is expecting, and hosting providers will have to support it regardless of the cost.

What does this mean for hosting providers? How can you prepare? Are we all going to go broke paying Comcast for IP addresses?

In parts one and two of this series, we’ve explored why the demand for IPv4 will remain strong for the foreseeable future. But in any market, both supply and demand are critical components. In the final part of this three-part series, we will explore why the idea of “IPv4 depletion” is misleading, as the supply of IPv4 addresses will be far greater for far longer than anyone assumes. Join us for part three to learn what to expect as a hosting provider in an IPv4-depleted world, and how to prepare yourself.