House of Eratosthenes

There are currently only 232 million IP addresses left — enough for about 340 days — thanks to the explosion in smartphones and other web-enabled devices.

“When the IPv4 protocol was developed 30 years ago, it seemed to be a reasonable attempt at providing enough addresses,” carrier relations manager at Australian internet service provider (ISP) Internode John Lindsay told the Herald.

“Bearing in mind that at that point personal computers didn’t really exist, the idea that mobile phones might want an IP address hadn’t occurred to anybody because mobile phones hadn’t been invented [and] the idea that air-conditioners and refrigerators might want them was utterly ludicrous.”

Nothing sells quite like the next Apocalypse.

Nothing sells quite like the next Apocalypse.

I teach CIS classes in my spare time. I also was a programmer during the Y2k “apocalypse”. I can almost guarantee you that the migration to IPv6 will proceed more or less painlessly because the preparations have been coming down the pike for a long time.

I’m trying to figure out what the next end times scenario will be. My guess is that it involves the Cubs winning the World Series.
- Physics Geek | 07/27/2010 @ 06:47
With luck, this will spur people into moving toward IPv6. I can’t wait to deploy OSPFv3 and BGP-6, get rid of DHCP servers altogether (since addressing is handled by the BIA on the card), and have enough available globally routable IP addresses to assign one to each cell in my body and have enough left over for every item I will ever touch in my life. Bring it, Internet!
- muttley | 07/27/2010 @ 08:11
Goodness… we’ve been running out of the internet since 1990.

(and IPv6 is a kludgy protocol that was not upwardly compatible with IPv4 which is why after 20 years it still hasn’t really gotten off the ground)

*sigh*
- pdwalker | 07/27/2010 @ 09:24
IPv4’s successor is not going to be upwardly compatible with IPv4, not if it in any way cracks the 4 billion address barrier. It is a mathematical constraint.

And the 4 billion is misleading, the number is substantially less than that. You’d have to review the Class A, B and C registrants to figure out where we stand with this, I think…but then you’d tack the other way as you add in the illegal networks.
- mkfreeberg | 07/27/2010 @ 09:38
When you figured that 16M addresses are wasted on the 127.0.0.0/8 block just for loopbacks, and then several other Class A’s are IANA reserved, that’s a huge bite out of it. Then there’s the class E networks which are “experimental” – who is experimenting with them? The multicast range is hardly used at all. That’s all the Class D’s. Then you get the large companies like GE, GM, Halliburton – they all have class A’s, and don’t bother running NAT, so they waste huge swaths of IP addresses, when they could use a few and then lease out the rest at a profit. (I’ve done lots of work on all of those networks, so I’m quite familiar with them.)

It’s a wonder we’ve lasted this long. RFC1918 bought us some time, but not an infinite amount.
- muttley | 07/27/2010 @ 09:44
“When the IPv4 protocol was developed 30 years ago, it seemed to be a reasonable attempt at providing enough addresses,” …

“640K is enough for anyone,” redux.
- bpenni | 07/27/2010 @ 10:02
That might be why they didn’t go into your level of detail, muttley, calculating the gross; it would emerge that 232 million available is not quite so close to the edge of the cliff after all. Something does have to be done, but that’s just the nature of allocating finite regions of memory, treating these arrays of on-off as integral numbers, and then building entire architectures around the agreed-on arrangement…which, when you get right down to it, is all software, firmware and hardware development really is.

Limits like these are encountered and then “broken” every day. It only becomes news because of the large number of people who will be impacted.
- mkfreeberg | 07/27/2010 @ 10:08
Transitioning should not be that difficult, however. There’s plenty of documentation on 6-to-4 tunnels, so people can go ahead and migrate their internal networks off of RFC1918 addresses and into IPv6. APNIC is already handing out IPv6 space to their people because APNIC is almost out of IPv4 ranges to give to people. For some DNS records, you can already see it returning an IPv6 entry. (For grins, perform a “dig” on sans.org, and look at the AAAA records.) The infrastructure is already in place to handle it. Windows and Linux have supported IPv6 for years, Cisco IOS has had it on almost everything for quite a long time. The holdup, sadly, is the backbone layer. If they start routing 6 across it, then the transition would be quite simple. Those slower to adopt could NAT their 4 addresses to 6 addresses quite easily (concatenate 0::/96 to the beginning of the address and represent it in hex instead of dotted decimal), and the problem is, indeed, solved. Just like the Y2K bug, the fix will be fairly painless for most people (‘cept us net engineers, but, hey, typewriter repairmen weren’t happy about the invention of the PC, either). The only thing I don’t like is that IPSec will be native to it, which makes it easier to bypass network-based IPS solutions.
- muttley | 07/27/2010 @ 10:37
[…] Morgan Freeberg Oh noes, we’re running out of internets! […]
- Cassy Fiano | 07/27/2010 @ 11:38
I’ve been hearing about IPv6 for years, and wondered why it still wasn’t common. Now I know, sort of.

Hey, how come there’s no IPv5? What happened to it?

Excellent points, muttley. I was able to follow MOST of what you’re saying, having worked on small networks and taken the first couple of classes for my CCNA. A little of it went over my head, but your reasoning is sound. Just who IS experimenting, and why DON’T class-A holders run NAT? Good questions.
- cylarz | 07/27/2010 @ 18:58

Leave a Reply

You must be logged in to post a comment.

Recent

Items of Interest

Monthly Archives

What They Say

Links

Running Out of InternetMonday, July 26th, 2010

Running Out of Internet
Monday, July 26th, 2010