Control of the Internet

The InternetIf you follow presidential politics you may have heard a few candidates claim that the US government is giving away control of the Internet. This one puzzled me, and it turns out what they are talking about the transition of the control of the DNS function from US control to a non-profit international body. It turns out that this is something that has been in the works for decades.

The issue involves DNS, or the Domain Name System. This is the system that matches the name of a web site with an IP address. This system allows you to go to the amazon.com website by typing the name address “amazon.com” into your browser instead of having to know the numerical IP address for Amazon.

DNS is essential to ISPs because it tells them how to route a given request on the web. There is one master file of all worldwide web names and the associated IP addresses. And obviously somebody has to be in charge of that directory to add, delete and make changes to web names and IP addresses.

After the early days of the Internet this function went to a group called IANA, the Internet Assigned Numbers Authority. This group was largely managed by a few staffers, academics, and help from some of the early web companies – all techies who only wanted to make sure that the burgeoning web worked well. And although they didn’t exert any control, the group was loosely under the auspices of the NTIA (National Telecommunications and Information Administration), a part of the Department of Commerce which had veto power over anything done by IANA.

This power was rarely exercised, but there were many around the world that were uncomfortable with the US Government being in charge of a vital web function. There was a push for an international group to take over the DNS function and in 1998 the function was transferred to ICANN, the Internet Corporation for Assigned Names and Numbers. ICANN brought in Board members from around the world and the group has effectively since then been operated with international consensus. But the NTIA still maintained a veto power over things done by the group.

But since it was founded there has been a planned transition to a fully international ICANN with no ties to the US government and on October 1 control of ICANN changed hands and is now operated only by an international Board without oversight from the US government.

Just a few weeks before the planned transfer four states sued to stop the transfer in the US District Court in Texas. Their argument was that the directory of IP names and addresses belonged to the US and could not be given away without approval from Congress.

The opponents to this suit argued that not turning over the control of ICANN was a much bigger threat because it might lead to other countries developing their own DNS databases – and the ability of anybody in the world to reach any web address using the same nomenclature is vital to the concept of an open and free Internet. Interestingly, it was this same concept a century ago – that anybody with a telephone ought to be able to call any other telephone number in the world – that was a driving principle in creating an efficient worldwide telephone network.

The suit was processed quickly and the judge came down on the side of the open Internet and the transition to ICANN. In the end this fight was more about politics than anything substantial. At the end of the day the DNS database is nothing more than the equivalent of a gigantic white pages listing of every address on the Internet. All that really matters is that this database be kept up to date and be available to the whole world. ICANN has had the same international board of techies since 1998 and this transition was planned for a long time. So there is no threat to the US losing control of the Internet folks that saw the headlines can sleep well knowing that this issue was about politics and not about a real threat.

The IP Address Crunch

4cb1f2dc96040Sometimes it feels like small ISPs just move from one crisis to another. The latest problem I am hearing about is that ISPs are having a hard time getting new IP addresses – which is something they need in order to connect new customers to their network. I have clients who have been trying for months to find new addresses, and if they don’t find any they are soon going to have to turn away new customers.

We’ve known for decades that we would exhaust the current IP addresses. The IP world introduced IPV6 IP addresses back in 2011 and that was supposed to be enough new IP addresses to last the whole world for a long time into the future. Historically the original Internet used IPV4 IP addresses, of which there was about 4.3 billion. The new addresses have more digits and there are about 79 with 28 zeroes after it times more IPV6 addresses. Even the tens of billions of expected IoT devises won’t make a dent in the new inventory of IP addresses.

So how can there be a shortfall of IP addresses with so many new ones available? The problem is the speed at which the world is implementing the new IPV6 addresses. Some of the large companies like Comcast, Verizon Wireless and T-Mobile have swapped all of their customers to IPV6 addresses. But the implementation has been slow. Google probably has the best measure of IPV6 implementation since they see a large chunk of the world’s traffic. By 2014 they reported that only 2% of the IP addresses in the world had been converted to IPV6. At the end of last month that had finally climbed to 14% of all IP addresses.

But so far the conversions have been done by the largest ISPs. It is exceedingly hard for small ISPs to make this transition. They are more or less locked into the IP practices of the large carriers that sell them Internet bandwidth. It’s been estimated that the small companies might not be offered IPV6 until perhaps 50% to 60% of the Internet traffic is using the new addressing standard. By the looks of the growth curve that is still at least a few years away.

The bodies that assign IP addresses have all run out of new addresses. The Internet Assigned Number Authority (IANA) free pool of numbers ran dry in February 2011. There are five Regional Internet Registries (RIRs) around the world and the last one of them ran out of IP addresses last year. Since then ISPs can’t get IP addresses through the normal channels.

So small ISPs are stuck in limbo. If they want to grow they need new IP addresses, but there are none available in the traditional channels. As happens with any scarce resource a new market of brokers has stepped in to supply the demand for IP addresses. There are several of these brokers worldwide. These brokers have gone to large companies like GE, Haliburton and Ford and bought their inventories of unused IP addresses. And this process created a market.

Back in 2012 these brokers established market prices for IP addresses. The prices started at about $5 per IP address. But as these brokers have found fewer unused blocks, and as there are more ISPs looking for numbers, the prices have risen and IP addresses today sell for between $11 and $15 per IP address.

So small ISPs should just be able to buy what they need from these brokers, right? Unfortunately it’s not that easy. The addresses are sold through a periodic online auction process, and like happens with any rare resource there are now speculators buying IP addresses with the hope of selling them later at a higher price. The competition in the auction processes has become fierce. To some extent this is like the process for trading bitcoins and those with the fastest and most powerful computers can win the auctions. The small ISPs I know tell me they are not getting any addresses. I know one ISP who has failed at the process for over 6 months.

So we now have a situation where small ISPs are nearly locked out of the process of buying new IP addresses (and even if they buy them they are expensive). This shortfall and the auction arbitrage is likely to last for a few more years. The economics of the market tell us that at some point the arbitrage price for IP addresses will drop. When that happens the speculators in the market will ditch their inventory and there should be IP addresses available at lower prices than today and more easily available. But that’s not expected until there are a lot more IPV6 users. The ISPs might be facing this problem for the next two years. I feel certain that we are going to see small ISPs that will find themselves unable to add new customers to their networks – and in world where we want broadband everywhere that is a disaster.

Finally Time to Convert to IPv6?

Rolling diceIn early 2011 the Internet Assigned Numbers Authority (IANA) allocated the last block of IPv4 addresses to various countries and warned that at the rate of historic usage that all those numbers would get gobbled up by ISPs within a few months. They further foresaw all sorts of new demands for IP addresses for new industry products like wearables, BYOD devices being connected to corporate WANs, an explosion of smartphones in the developing world and the early stages of the Internet of Things.

The IANA warned then that ISPs should begin migrating to IPv6 to avoid running out of IP addresses. But here we are almost four years later and a lot of ISPs still have not converted to IPv6. And yet somehow we are not quite out of numbers. The American Registry for Internet Numbers (ARIN) for the United States and Canada is still handing out IP addresses even today. How is that possible?

The major Internet players in the markets developed ways to conserve and reclaim IP addresses. At the end of 2013 ARIN still had 24 million addresses available. ARIN has been able to stretch the numbers by doing things like reclaiming IP addresses from dead ISPs, and by doling out IP addresses in much smaller blocks than historically. ARIN was predicting that those addresses would be gone before the end of 2014. But again the industry confounded them and there were still 16 million IP addresses at the end of 2014. You can see the count of available addresses at this web site, which is updated weekly.

So is now finally the time to convert to IPv6 or can the industry stretch this further? The issue is going to be of the most concern for growing networks that need a lot of new addresses. If you are growing you might should  convert to IPv6 before you find yourself stopped dead due to lack of IP addresses. ISPs that are not growing are probably good for some time since they can usually use numbers abandoned by old customers and assign them to new ones.

Why haven’t more ISPs converted to IPv6? There are a number of reasons.

  • IPv6 is not backwards compatible and once you convert you need to run what is called a dual stack that will process both IPv4 and IPv6 addresses. And you will have to do that until IPv4 addresses are finally dead.
  • The conversion to IPv6 can be expensive. Every part of your network needs to be IPv6 compatible – that means core hardware, end-user hardware, your Internet backbone provider and even content providers on the web.
  • Not all of the content on the Internet is IPv6 compatible. Almost every major site like Google, Yahoo, Facebook and anybody else with a household name is now IPv6 compatible, but there are older content providers who have not bothered, and maybe who will never bother to make the conversion. (And it’s not so much the content, but the servers they sit on).
  • There is no problem buying new hardware that is IPv6 compatible. So any new gear being installed today is already IPv6 compatible. But this is no help for older routers and gear that is not easily upgraded, and many companies are holding off to avoid the capital outlay or upgrades that will be needed to convert to IPv6. There probably is no way to upgrade a 10-year old DSL modem or a DOCSIS 1.1 modem to IPv6 without changing devices.

But we are finally seeing some large players converting. For example, Comcast converted all of their cable modem customers to IPv6 during 2014. But some other large ISPs are still holding back. Other parts of the industry have converted – most smartphones now use IPv6 as well as new game consoles.

Now that some large ISPs have converted and most of the web has converted most industry experts expect the rate of conversion to accelerate. Small ISPs need to pay attention to what is happening with IPv6 because you don’t want to be the last one to make the conversion. Once most of the rest of the world has converted you can expect to start having compatibility problems in unexpected places. And eventually the large carriers are going to declare IPv4 dead and cut it off. I’m thinking that most small ISPs ought to finally think about converting by no later than next year. If you wait longer than that it becomes a crap shoot.