Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×
Network The Internet Networking

The RIPE NCC Has Run Out of IPv4 Addresses (ripe.net) 172

Kelerei writes: The RIPE NCC, the regional internet registry for Europe, West Asia, and the former USSR, has allocated the final /22 remaining in their address pool, and has stated that they have now run out of IPv4 addresses. RIPE will continue to recover IPv4 addresses from organisations that go out of business, close, or that return them due to lack of need, but expects these small amounts of recovered IPv4 addresses to fall well short of demand: RIPE will now only assign IPv4 addresses to entities that have never received any IPv4 allocation in the past, and even then will assign no more than a /24. RIPE puts out a call to action for IPv6 migration:

This event is another step on the path towards global exhaustion of the remaining IPv4 addressing space. In recent years, we have seen the emergence of an IPv4 transfer market and greater use of Carrier Grade Network Address Translation (CGNAT) in our region. There are costs and trade-offs with both approaches and neither one solves the underlying problem, which is that there are not enough IPv4 addresses for everyone. Without wide-scale IPv6 deployment, we risk heading into a future where the growth of our Internet is unnecessarily limited — not by a lack of skilled network engineers, technical equipment or investment -- but by a shortage of unique network identifiers. There is still a long way to go, and we call on all stakeholders to play their role in supporting the IPv6 roll-out.


This discussion has been archived. No new comments can be posted.

The RIPE NCC Has Run Out of IPv4 Addresses

Comments Filter:
  • by Gregory Eschbacher ( 2878609 ) on Monday November 25, 2019 @02:07PM (#59453072)

    As with any scarce resource, it can never really just "run out". Instead, as it becomes increasingly scarce, the market will react to changes in the elasticity curve.

    In this case, as easily obtainable IPV4 addresses run out, the price should go up. As prices go up, organizations that have large blocks of IPV4 addresses could sell them which increases the overall supply. Older companies (HP? IBM? ) from the good ol' days of computing are probably holding onto vastly more blocks of addresses than they could ever need and would be able to sell them. Organizations (universities?) that have large blocks of IP addresses could also move to put themselves behind NAT or equivalent technologies and free up addresses.

    I think there really isn't much to worry about here, especially since we've been talking about this since 2011 (https://tech.slashdot.org/story/11/04/14/2237237/asia-runs-out-of-ipv4-addresses ) and 2015 (https://tech.slashdot.org/story/15/07/02/1437236/north-america-runs-out-of-ipv4-addresses ) and the world just keeps on moving even as the number of connected devices has exploded. How? Because efficient markets in allocating scarce resources have come into effect.

  • by JcMorin ( 930466 ) on Monday November 25, 2019 @02:15PM (#59453090)
    What blow my mind is how many /8 the departement of defense in the us have. I can't believe more than 1% of those are in used. https://en.wikipedia.org/wiki/... [wikipedia.org]
    • Re: (Score:3, Interesting)

      by Anonymous Coward

      What blow my mind is how many /8 the departement of defense in the us have

      The reason those were issued is clear. It's the reason they haven't released them back that's sucky.

      The DOD and some universities were the original arpanet infrastructure, and this was a time before CDIR subnetting existed.
      The "netmask" of an IP block was defined by the highest 4 bits of the IP, and those bits being all zeros made it an "A class", or what we now call a /8
      This is also why 0.x up to 127.x are listed the way they are, those were all A-class and at the top level couldn't be divided. 128.x was

      • Please upvote this anon. They are right for the issue reason. Before CIDR (the parent fat-fingered the acronym), a class "A" was a class "A". Once your first octet started with an 0-127, it would be treated like a /8 by all hardware and software at the time. So giving it out was all or nothing, there wasn't a way to be smart about it.
      • I can think of a reason. They're valuable, and they're increasing in value every day. There are a lot of reasons the US government would want to hold onto a limited resource that constantly increases in value. Whether that's ethical or not is another question, of course.
  • From an ISP's perspective:

    A. Put customers behind CGNAT, charge extra for a public IPv4 address.
    B. Give everyone a block of IPv6 addresses for free.

    I know some ISPs have chosen B (Comcast, surprisingly) but I would be willing to bet that a not insignificant chunk of ISPs are going to pick A for as long as they can get away with it. (That is, until a critical mass of "important" online services go IPv6-only. Which, let's face it, that's going to be a while.)

    • Doesn't being behind a CGNAT help protect your privacy a little?

      • Technically? Nothing really stopping the ISPs from selling the customer-to-port mappings to advertisers, but it *does* add a step to the process.

        That's probably how they'll sell it, though.

        New feature: "Privacy Guard CGNAT" protects your privacy by preventing your computer from broadcasting an IP address. Please note that some applications such as VPNs and certain games are not compatible with "Privacy Guard CGNAT." Public IP addresses are available for the low cost of $29.99/month for our valued customers

    • As soon as one porn site goes IPv6 only, the internet world will change overnight.

      • Well, yeah. Why do you think I put "important" in quotes?

        You didn't think I was talking about the Smithsonian or something, did you?

    • At my ISP (Aussie Broadband):

      A. Customers are put behind CGNAT by default. If a customer needs a public IPv4 address (e.g. for gaming, to run a server, etc) they can get a sticky public IPv4 address for free (easy to request). Approx 10% of customers are requesting public IP addresses.

      Note: a sticky IPv4 address is not guaranteed to remain the same over time, but in practice almost never changes except when there are major changes in the network e.g. when CGNAT was brought in.

      B. A static IPv4 address is ava

  • At this point it has become clear that market forces alone are not enough to move us all to IPv6. Free-market is great when it works, but sometimes it needs a little push, some correction, from above. I think that IPv6 should be promoted by states. How much? I don't know. That can go from tax benefits to just making it a mandatory thing ISPs must implement (same as governments mandate standards in other fields).
    • by Bert64 ( 520050 )

      Well the regional registries allocate addresses to their members and require their members comply with various rules... So years ago they should have required that their members must provide dual stack or ipv6-only services as a requirement of getting or maintaining an address allocation. In fact, start taking back ipv4 allocations from any organisation that's providing ipv4-only services unless they provide dual stack by default for all their services.

      Another one is demand and awareness. If google and othe

    • by jrumney ( 197329 )

      I switched over to IPv6 for everything 3 weeks ago when my ISP started using CG-NAT instead of publically accessable dynamic IPs. I still have outgoing IPv4, but I've updated my AWS instances so I can use native IPv6 between there and my home. The difficulty I'm having now is that dynamic DNS clients and providers seem to make it hard not to set an IPv4 address. I figured out how to do it manually with DuckDNS, but still trying to get the client to play ball. If my ISP would assign a static IPv6 prefix,

  • by thatseattleguy ( 897282 ) on Monday November 25, 2019 @02:31PM (#59453138) Homepage
    Just curious. The fine page at: https://en.wikipedia.org/wiki/... [wikipedia.org]

    ....shows 240.0.0.0/8–255.0.0.0/8 (aka 240.0.0.0/4 - a full 6.25% of the available IPv4 address space) - as "reserved for future use". That's apparently for IP Multicasting per various RFCs (919, 922).

    Could those conceivably be made available, or is 240.x.x.x so deeply enshrined in various TCP stacks as "multicast only" that it can never actually become available for general-purpose usage?

    • Why bother risking problems ? That extra 6% only buys people a little more time to be lazy and do nothing. Everything is ready for IPv6.

      • "Everything is ready for IPv6."

        Except you know, people.

        Its an important part of the puzzle. Take a look at an ipv6 address, it makes no sense. I am sure there is some way to read them but i dont use them enough to learn the syntax. Catch 22 i guess.

        • Makes perfect sense...
          First two parts is your prefix eg 2001:2020:, which identifies your company.
          Then how you allocate the addresses underneath that is up to you... Most will come up with a scheme based on site or vlan ids etc.
          2001:2020:100:: - site 100 (hq)
          2001:2020:200:: - site 200 (backup dc) etc
          2001:2020:200:10:: - site 200, vlan 10
          2001:2020:200:10::50 - site 200, vlan 10, host 50

          You don't have to use the autoconf based on mac address, but you can if you want.

          It's much easier than the mess which is ipv4... Where does 18.3.1.32 sit on the network in relation to 64.21.2.5? And when you connect to port 80 of 18.3.1.32 does that forward to 192.168.10.10 or 10.2.3.50?

          I find the ipv6 MUCH easier to manage and remember than the legacy v4, where i have to maintain a database to know where a given subnet or host is, with disparate address blocks all over the place and then nat adding extra confusion.

          On the other hand i only have to know this because i manage a network... For the average user, they just use DNS.

    • by Z00L00K ( 682162 )

      Multicast is used for a variety of functions, among them are routing protocols and without multicast the routing won't work.

    • by higuita ( 129722 )

      there is a project reassigning those reserved IPs to a new use and most of the last few years IP allocations are being done using those reserved IPs.
      The problem is some of then are in use or blocked and may require firmware/config updated in routers.
      notice that even the 0/8 block is being take care, so it may be possible to allocate a 0.1.2.3 address in the future
      of course, the energy put in doing this could also be used in pushing the ISP to finish their IPV6 migrations. With ipv6 solved in the ISP, everyt

    • by jrumney ( 197329 )

      Multicast is used by mDNS and UPnP for discovery. There also used to be video services using it though it fell out of favour for that because of routing issues when not all routers are configured for multicast (and for ISPs, the routing tables to handle it can get quite big when a lot of customers tune in to multicasts at once, so a lot of them turn it off).

  • by Greyfox ( 87712 ) on Monday November 25, 2019 @02:57PM (#59453262) Homepage Journal
    Elon's putting a few tens of thousands of internet-providing satellites in orbit, he could just make them ipv6 only and that would probably pretty much single-handedly drive adoption of the standard. Otherwise he might have to whore himself out to IBM or someone -- IIRC those guys got three class As back in the day and they just use them on the internal corporate network.
    • A service which so far only counts Elon Musk as a user doesn't exactly carry much clout. What you need is either a consortium of to the top ISP's (ATT, Verizon, Spectrum, etc) or a consortium of the top services providers (Google, Facebook, Amazon, Netflix) giving the other a deadline by which time ipv4 will be turned off. Of course it would be nothing more than a bluff since neither side is going to risk pissing-off users and loosing business over ipv6
  • I run a small-ish AWS rollout with about 55 hosts. I've enabled IPv6 on all of them. But there are still plenty of clients connecting w/ IPv4, so I can't really remove the IPv4 addresses yet. It's kind of annoying, because AWS is much nicer with non-transient IPv6 addresses than it is with IPv4 addresses that change every time an instance is stopped. People gotta get their shit together.

  • Relevant XKCD:
    https://xkcd.com/865/ [xkcd.com]

  • by PPH ( 736903 ) on Monday November 25, 2019 @03:47PM (#59453524)

    Once the UK drops out of Europe and gives all their IP blocks back, there should be plenty for the remaining members.

  • Welcome to the new era of widespread adoption of IPv4 CGNAT. Forget using torrent or any domestic server: cloud-based (and subscription-based) services only, sorry, your old RaspberryPi-based appliance no longer works in our network.. The IPv4 end-to-end principle, long ago deadly hurt by NAT, will definitely dissapear.

    Time to select an ISP providing IPv6 service (none in my area, though) and to consider throwing away all those IPv4-only appliances...

    • that's kind of alarmist nonsense considering ISP have huge pools of ipv4 for the end user. this impacts mostly nobody as a customer.

      • by Bert64 ( 520050 )

        Only in some places...

        https://en.wikipedia.org/wiki/... [wikipedia.org]

        Some countries are screwed, and only have a single ipv4 address per several thousand of their population. North korea has only 1024 addresses in total, which is fine if only the dear leader is allowed internet access but for them to ever offer any kind of service to the population they would have no choice but to use cgnat.

        Every isp in myanmar puts all their customers behind cgnat, some will provide a dedicated ipv4 address to business customers for a l

  • It's not like we need more anti-democracy bots

  • It's a shame they didn't use 16bits (IPv2?), so that by the time big greedy corporate types caught on they would not have been allowed in.

    But really, just keep dragging your feet on ipv6 ISPs...

Truly simple systems... require infinite testing. -- Norman Augustine

Working...