The IPv4 Sky Really Isn't Falling - InformationWeek

InformationWeek is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them.Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

IoT
IoT
IT Leadership // Digital Business
Commentary
11/10/2014
03:00 PM
Lawrence Garvin
Lawrence Garvin
Commentary
100%
0%

The IPv4 Sky Really Isn't Falling

The shortage of IPv4 addresses has more to do with misallocation than scarcity.

We've been hearing for years that we're running out of IPv4 addresses. But we're not running out -- they're not all in use. It's just that some organizations have no more addresses to allocate to subordinate organizations. More significantly, I submit that this is an artificial shortage of addresses caused more by the mismanagement of their allocation than by scarcity.

There are 221 groups, each with about 16.7 million addresses available for assignment. There are certainly a lot of devices available via the Internet -- but are there more than 3.7 billion of them? I'm not talking about devices used to access the Internet. I'm specifically concerned only with those devices directly reachable via the Internet, such as servers (web, mail, FTP, DNS, reverse proxy), routers, and firewalls. As it happens, this is a difficult question to answer because everybody is busy counting devices with access, not devices that only provide services.

But let's set aside the raw numerical comparisons and get back to the real issue. The issue isn't that all the IP addresses are in use; it's that all of them have been allocated. Many of those allocations are grossly wasteful, and millions of addresses are unused. A less kind person might even say they're being hoarded.

[The Internet of Things will require a balance between disruptive and incremental change. See Is 1% Improvement Boring, Or A Breakthrough?]

Let's review how IP addresses get allocated. At the top, the Internet Assigned Numbers Authority (IANA) allocates blocks of addresses to the Regional Internet Registries (RIRs), of which there are five: ARIN (North America), LACNIC (Latin America), RIPE (Europe, Middle East, Central Asia), APNIC (Asia/Pacific), and AFRINIC (Africa). IANA has allocated all available blocks to those five RIRs, which then allocate smaller blocks of addresses to ISPs, government agencies, and organizations. Of particular note, APNIC allocated all of its addresses by April 2011. As many as 350 million addresses were allocated to China, ostensibly in use behind the national firewall that nobody can get in or out of. I'm not aware that any of the other four RIRs have reported allocation of all available addresses.

Let's focus on ARIN and North America, because that's where most of the waste originates. Most of the allocations in North America occurred prior to the introduction of Classless Interdomain Routing (CIDR) in 1993. Allocations made after the introduction of CIDR were done in much smaller blocks and thus typically more efficiently.

For the sake of this argument, I'll focus only on the pre-1993 "Class A" address pool, which was allocated in blocks of 16.7 million addresses at a time and is where reforms could yield the most immediate gains.

  • Did you know, for example, that only five of the 16.7 million addresses allocated to the US Postal Service in 1992 appear to be visible publicly? Three DNS servers and two SMTP servers. The website is hosted elsewhere.
  • Did you know that 11 Class A networks (as many as 183 million addresses) are allocated to the US Department of Defense Network Information Center? How many of its systems even have Internet access, much less provide Internet services?
  • Did you know that more than a dozen US corporations have Class A network space allocated to them? That's more than 16 million addresses per corporation. Among them, according to Wikipedia, are Apple, DuPont, Eli Lilly, GE, Hewlett-Packard, IBM, Merck, and Xerox.

To their credit, some organizations have been voluntarily returning their over-allocated addresses to be reissued where needed. Kudos to Stanford University, for example, for releasing the 36.0.0.0/8 network (which was then given to APNIC and is now allocated to CHINANET).

What to do?
In that short analysis, I've identified about 400 million IP addresses just in the US that likely could be reclaimed in short order. And I suspect that most of the organizations with those 16.7 million addresses aren't using most of their addresses and could easily give them up.

So, yes, without any change it's quite likely that at some point ISPs will no longer have any of those 3.7 billion IPv4 addresses to provide to customers. Is it time to implement some changes from the top? Is your company using public IP addresses inside its networks where it could be using private IP addresses? Do you have the capability to properly assess and manage your IP address usage? Will handing IANA over to an international governing body have an impact? Please tell us what you think in the comments section below.

Apply now for the 2015 InformationWeek Elite 100, which recognizes the most innovative users of technology to advance a company's business goals. Winners will be recognized at the InformationWeek Conference, April 27-28, 2015, at the Mandalay Bay in Las Vegas. Application period ends Jan. 16, 2015.

Lawrence Garvin, head geek and technical product marketing manager at SolarWinds, wrote his first computer program, in RPG-II, in 1974, to calculate quadratic equations. He tested it on some spare weekend cycles on an IBM System 3 that he "borrowed" from his father's ... View Full Bio
We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.
Comment  | 
Print  | 
More Insights
Comments
Newest First  |  Oldest First  |  Threaded View
MikeB702
50%
50%
MikeB702,
User Rank: Apprentice
11/14/2014 | 8:30:07 PM
Re: Observations on the IPv4 Falling Sky
IPv6, then called IPng for next generation, got off the ground right around the time that the Internet first appeared poised to run out of addresses in the early 1990s. Two major change took place around the same time, the introduction variable sized subnets (other than the inefficient original class A, B, and C) and the introduction of NAT. The effects of both of these changes was to reduce the speed of IPv4 exhaust, and IPv4 exhaust was the only reason for IPv6.  NAT was always the redheaded stepchild of the Internet, and early on it had many problems. But over time the Internet organism got used to NAT and developed workarounds for the most common problems. But the designers and pushers of IPv6 disregarded NAT as the essential element it became, and kept flogging the IPv6 option. Long ago we were supposed to have transitioned naturally through the use of dual-stack, but there was always a lack of economic incentive to even take the half-way step of dual-stacking. Now we have reached (nearly) IPv4 exhaust and IPv6 is nowhere near ready to take over. So we have entered the IPv4 market era, where unused addresses can be sold to those who need them.

The technorati always disregarded NAT, viewing it as degraded Internet or otherwise a technically inferior product. Their assumption was that the natural superiority of IPv6 would provide all the incentive necessary for a timely transition, ignoring the fact that the Internet responds to economic incentives which are completely missing from the IPv6 situation. Now we have ISPs who offer NATted addresses as the default, but upon request will issue a real one. Those requests are few and far between, though.
Sherly Mendoza
50%
50%
Sherly Mendoza,
User Rank: Apprentice
11/14/2014 | 7:32:03 PM
Re: Observations on the IPv4 Falling Sky
So why all the impending level of doom we have been hearing for a ver long time? Did the forecasters ever include factors such as the NAT? Was the behavioir of companies considered? In the end it does look like this is simply following the pattern in the tech industry. There are several examples where new standards are intoduced before old ones reach their full usefulness, hence they end up running side to side.
MikeB702
50%
50%
MikeB702,
User Rank: Apprentice
11/13/2014 | 10:45:00 AM
Observations on the IPv4 Falling Sky
I liked the article. The author senses the distinction between devices which need a routable public IP address and the far larger number of devices which do not. According to Cisco there are already 10 billion devices connected to the Internet. More than the number of  humans, more than the number of IPv4 addresses. Yet the Internet functions well every day. Where is the pressure to move this behemoth to IPv6? The word that is missing from the article is NAT. IPv4 NAT at multiple levels allows us to smoothly and transparently use IPv4 addresses to their fullest capacity. This is because Internet communications flow not from unique IPv4 address to unique IPv4 address. They flow in sessions between unique IPv4 address + TCP Port number combinations. Fortunately each unique address supports 64,000 TCP ports (plus another 64,000 UDP). We go crazy ramping up from 32 to 128bits for network addresses and yet ignore the obvious benefits of the extra 16 bits in the IPv4 header that functionally expand address space enormously. You can point out some problems with NAT and some benefits from a flat IPv6 space, but those don't add up to any real incentive for IPv6 transition.

More than the large number of allocated-but-not-advertised or advertised-but-not-utilized addresses, there is the enormous multiplicative advantage of multi-level NAT deployment that will serve to keep the IPv4 sky from falling. For the very rare cases where a real routable IPv4 address is needed, it can be supplied on the same infrastructure that is supplying NATted addresses to the 99% of users who don't care. Everybody is happy, the 99% whose Facebook still works, the 1% with the static IPv4, and the carrier who has leveraged his IPv4 holdings.
SaneIT
50%
50%
SaneIT,
User Rank: Ninja
11/12/2014 | 7:42:06 AM
Re: Brother can you spare an IP address
I understand that /8 networks were mentioned specifically but most of the owners of those blocks seem to be land grab type assignments and they are most likely being used exactly like those /24 networks that I own.  This isn't a matter of ISPs holding /8 networks to assign public IP addresses to customers, after the US military takes 1/3 of the blocks most of them are private companies who will never use 16 million addresses.  In the end this still boils down to over subscribing because rules of years gone by can still affect how things are used today.  
Sherly Mendoza
50%
50%
Sherly Mendoza,
User Rank: Apprentice
11/11/2014 | 10:37:36 PM
Over Preparedness
We cannot blame the entire IT industry for becoming over prepared i guess. "Eventually" all the addresses will be used or perhaps allocated anyway, but as demonstrated in this article it is actually more of a matter of time. A long time. What I am surprised to find out is that there are many cases where groups grossly over projected their need and then just let it flow in aether. If it were that important, government oversight committees would be breathing down their necks by now.
ramsfan93
50%
50%
ramsfan93,
User Rank: Apprentice
11/11/2014 | 12:14:47 PM
Re: Hoarding
"v6 is inevitable".  OK, I first heard that in 1996.  

Yes, IPv6 may eventually replace IPv4, but it won't happen in any reasonable timeframe.  Do right by your business by using mature, reliable technologies, until something better comes along.  IPv6 is not there yet.

Oh yea, in 1996 I also heard that token-ring replacing ethernet is inevitable, so start replacing... oh crap wait. Forget it.
Lorna Garey
50%
50%
Lorna Garey,
User Rank: Author
11/11/2014 | 9:39:19 AM
Re: Hoarding
Exactly, Joe. Instead of running around trying to scrape together IPv4 addys, just move on -- v6 is inevitable, bite the bullet now before the business is impacted. 
LawrenceGarvin
50%
50%
LawrenceGarvin,
User Rank: Author
11/11/2014 | 7:24:46 AM
Re: Brother can you spare an IP address
Totally understand about the considerations for BGP, except that BGP has supported CIDR since 2006, and the article was about /8 blocks, not /24 blocks. :)
SaneIT
50%
50%
SaneIT,
User Rank: Ninja
11/11/2014 | 7:13:28 AM
Brother can you spare an IP address
On one hand I do know that there are many organizations out there who have more IP addresses than they will ever use but on the other I know why they have those addresses.  If a smallish company who will only have a handful of public facing IPs is making their living via the public facing services they are going to follow standard protocol and implement BGP so that they can spread some risk.  The problem there is that many ISPs will require a /24 network to advertise BGP so the company that needed 10 public facing addresses now has at least 254 and more likely 508+ addresses.  This cascades over time, I can log into my ARIN account and see that I still "own" addresses from a job I was at 20 years ago and that company is long gone.  
Joe Stanganelli
50%
50%
Joe Stanganelli,
User Rank: Author
11/11/2014 | 6:42:03 AM
Hoarding
It's only natural.  You tell people that there's a shortage of IPv4 addresses--of COURSE those with means are going to hoard them!

Nonetheless, even without the hoarding, broad IPv6 deployment needs to happen if Cisco's predictions about IoE are going to come true.
Slideshows
10 RPA Vendors to Watch
Jessica Davis, Senior Editor, Enterprise Apps,  8/20/2019
Commentary
Enterprise Guide to Digital Transformation
Cathleen Gagne, Managing Editor, InformationWeek,  8/13/2019
Slideshows
IT Careers: How to Get a Job as a Site Reliability Engineer
Cynthia Harvey, Freelance Journalist, InformationWeek,  7/31/2019
White Papers
Register for InformationWeek Newsletters
Video
Current Issue
Data Science and AI in the Fast Lane
This IT Trend Report will help you gain insight into how quickly and dramatically data science is influencing how enterprises are managed and where they will derive business success. Read the report today!
Slideshows
Flash Poll