3

Where do I get information on the IPv4 crunch (what will happen, transition to IPv6, etc.)? What should one do to prepare (if anything)?

Ars Technica ran "A decade's worth of IPv4 addresses" and reminded me about this.

Peter Mortensen
  • 2,318
  • 5
  • 23
  • 24
Nathaniel
  • 135
  • 5
  • 3
    We are going to run out of IPV4 address space in 18 months ... At least that is what they said in 2000, 2002, 2004, 2006,2008 ... and now ... 2010. – Zypher Jan 05 '10 at 06:37
  • 1
    We are going to run out of IPv4 addresses. The address space is finite and it is used at a finite rate. So, the people who claim it can go on forever need to check their maths seriously... – bortzmeyer Jan 12 '10 at 14:56
  • @bortzmeyer - NAT, 'nuff said – MDMarra Jan 16 '10 at 23:12
  • 4
    NAT is an awful hack, which creates a lot of work for the poor programmer and makes the Internet more brittle. In the end, the only question which matters is: who pays? With NAT, programmers and users pay... – bortzmeyer Jan 18 '10 at 07:49

5 Answers5

4

The best source, by far, for the future of IPv4, is Geoff Huston's IPv4 Address Report.

If you want to follow the news day by day, see the IPv4 depletion site.

bortzmeyer
  • 3,941
  • 1
  • 21
  • 24
2

I am a bit of a contrarian ... I am very leery of IPv6. There are privacy implications, because part of the IPv6 address is a unique device identifier, although I believe there are settings to make that random.

I fail to see the urgency, when so many class A networks are allocated to organizations with relatively minuscule need. To name a few: MIT, Daimler Benz, Xerox PARC, DuPont, Merck, the US Postal Service, Nortel, Eli Lilly and Halliburton. Heck, the InterOp show has a Class A for itself!

Back to the question. If you haven't already, read the Server Fault IPv6 tagged questions.

I don't think IPv6 is a huge issue right now for internal/private networks, but some preparatory steps are advisable for businesses. Eyes and ears should stay open, and the subject revisited every budget season.

  • I would make sure my internet facing hardware can support IPv6, or make sure I have sufficient budget to buy replacements in the event an internet provider switches.
  • I would also make sure any new networking equipment purchased can support IPv6 and IPv4.
  • I would find the networking equipment that doesn't support IPv6, and determine what needs to be done (firmware, upgrade, replace) for each item in the event IPv6 is thrust upon you.

I don't see that a homeowner/hobbyist needs to do anything. If your ISP switches, you will need a new modem/router, but you will have plenty of notice.

Community
  • 1
tomjedrz
  • 5,974
  • 1
  • 16
  • 26
  • 7
    Even if IANA can retrieve the /8s from organisations who have them unnecessarily (and really, would you want to renumber MIT so that it isn't spread out across it's /8?), that wouldn't add much to the global pool as some god-awfully huge quantity of addresses is being assigned each week from the RIRs. So it's a small and temporary stop-gap measure at best. – womble Jan 04 '10 at 22:59
  • 1
    Perhaps .. I just hear "the sky is falling" in so many of these arguments. How many companies out there are using even 50% of their allocation? – tomjedrz Jan 04 '10 at 23:39
  • In terms of "allocated subnets" within the organisation, they're probably pretty well utilised, but the allocations will probably be sparsely utilised. But even if they're not, *it doesn't matter*; even if all the unused addresses were reclaimed, it still wouldn't put a meaningful dent in the available pool, and the cost of reclaiming that address space is large, and would be borne entirely by the organisations which need to give up the resources. If I asked you to spend $200,000 so that you could give me half your backyard to be used for a public park, would that seem like a good deal? – womble Jan 04 '10 at 23:48
  • 2
    I see your point about the underutilized class A networks, but in reality if all the companies you listed gave back their entire network it would only put off the inevitable for a few months (IP4 address allocation was well over 200M/year last time I checked). Seriously, people have known about this problem for twenty years and the solution - IP6 - has been around for 12 years so it's not like an extra few months will really help anything. – phoenix8 Jan 05 '10 at 02:47
  • There are some issues at play that drive economists crazy. 1: you can get IP addresses by demonstrating need, so comcast uses an address for each set top box and phone and ... so they "need" 10 billion addresses. 2: you can't "sell" your addresses, so there is no incentive for MIT to want to sell their resource. If they could sell some /12s, I'm sure they'd find the resources to reIP their /8. Same especially goes for nortel and many of the rest. It's a real mess – chris Jan 12 '10 at 15:22
  • @tomjerdz The IPv4 sky isn't falling, but it will likely get more expensive. If I were an HP, Ford, or Halliburton investor I'd be ticked off if they didn't start leasing chunks of their /8s within the next 24 months. – Gerald Combs Jan 12 '10 at 17:12
  • @Gerald Combs Especially HP, who have three /8s - second only to the US government, AFAIK. – Richard Gadsden Apr 12 '11 at 14:48
  • @tomjerdz: And how many companies (last mile ISPs in particular) are using something like 800% of their allocation, thanks to NAT? – Ernie Jun 09 '11 at 18:17
  • @Ernie .. pretty much every company connected to the internet. Which is good. This is relevant to the discussion how? This is what we WANT. Personally, I think separating private networks from the public internet is good practice anyway. – tomjedrz Jun 10 '11 at 13:44
1

There is also a lot of useful information on the IPv6 Act Now site.

Jakob Borg
  • 1,453
  • 1
  • 10
  • 13
1

There are in fact a few questions on Server Fault already about IPv6 adoption and allocation of IPv4. Take a quick look at questions tagged IPv4 for example. There are a lot of 'technical' questions about IPv6 but there are also a lot of rollout questions, and people asking similar things to what you've asked.

Community
  • 1
Mark Henderson
  • 68,823
  • 31
  • 180
  • 259
1

There's a good discussion of IPv4 Address Exhaustion on Wikipedia.

I believe RFC 4941 addresses the privacy concerns that tomjedrz mentioned regarding IPv6.

Teddy
  • 5,204
  • 1
  • 23
  • 27
Dennis Williamson
  • 62,149
  • 16
  • 116
  • 151