Slashdot Mirror
What Happens When IPv4 Address Space Is Gone
darthcamaro writes 'We all know that IPv4 address space is almost all gone — but how will we know when the exact date is? And what will happen that day? In a new report, ARIN's CIO explains exactly what will happen on that last day of IPv4 address availability: '"We will run out of IPv4 address space and the real difficult part is that there is no flag date. It's a real moving date based on demand and the amount of address space we can reclaim from organizations," Jimmerson told InternetNews.com. "If things continue they way they have, ARIN will for the very first time, sometime between the middle and end of next year, receive a request for IPv4 address space that is justified and meets the policy. However, ARIN won't have the address space. So we'll have to say no for the very first time."'
You conveniently cut out the part of the quote that said ARIN would "receive a request for IPv4 address space that is justified and meets the policy". Have you ever applied for IPv4 space? ARIN does say no if your application does not have sufficient justification. I've had it happen, when someone decided we needed to apply for space when we hadn't really filled our existing space (it was just assigned inefficiently).
Comment removed based on user account deletion
Ok, let's say the IPv4 space ran out today and your ISP said you now have to run your server out of an IPv6 address.
You're now forced to move your server to another ISP that still has addresses available (probably ones that will start NATing all their non-server based clients so they can use their IPv4 allocations for server use).
If ISPs start moving non-server clients over to IPv6, then things will transition slowly, and at some point (ie. in 5 years) it will become feasible to run a server solely in the IPv6 address space as it will be accessible by the majority of users. Things progress this way until only a few dedicated IPv4 servers/clients are now safely behind translation routers.
However, instead of using IPv6, the sad thing is those ISPs will probably use IPv4 NAT to do the translation. The net effect is we push the crunch out a couple more years, but the following future is likely to develop as:
Fast forward a couple of years and now you find that all the ISPs charge a significant amount extra to run your server from an IPv4 address. You just pay more as it's just business as usual and you have no other choice. The ISPs with huge allocations are all laughing as they can leverage their allocated spaces at ever increasing dollar amounts. It's wonderful! The geeks aren't happy, because now it costs a lot more money to run their non-profit servers. Big business doesn't care, because it helps them by increasing the barrier to entry for smaller companies trying to compete with them on the internet front.
Fast forward five more years and things are now getting out of hand. Everyone is running behind NATed 10.x.x.x addresses (except large public servers), every second URL contains a port designator, port 80 web servers are now a luxury, ISPs are giving users the option of cheaper port redirects back to their own servers, and people are claiming that we've solved the problem for another 10 years.
Still the geeks are worried, but no one else cares. They now have less 'cruft' on the internet to worry about, and as long as they can still get to their Bittorrent/Porn/Facebook/YouTube they are happy as Larry.
The Class A owners will sell off chunks of their space one B class at a time.
...and we can watch the nerds scramble to upgrade their home and work enterprises so they can access it. :-P
I'm joking, or at least I think I am. If Slashdot did that I'm sure I would put more effort into getting an ipv6 address.
I was part of Open Systems Interconnection, OSI. We were pushing one of those many technologies like XNS, CHAOSnet, DECnet, IPX, SNA, and ATM/SONET that 'competed' with TCP/IP (NCP had been beaten back by then;^). Before the days of NAT, I had a "very persuasive" presentation that showed the Internet running out of 32-bit IP addresses by 1995 (China and India were my big closers that silenced a lot of TCP wonks). OSI had a 'better' addressing scheme that did everything -- distinguished end systems (ES) from intermediate systems (IS), facilitated class of service, extended addressing to the transport/session/presentation layer services, incorporated MAC layer addressing, facilitated source routing, provided network management hooks, and would give you a blow job that pealed the cover off a plenum cable. It was the ultimate networking addressing scheme. The routing vendors, who were accustomed to shoving the whole network layer address into a 32-bit register, said they couldn't implement a 20+ byte NSAP address, even though they only had to route on a small portion of it. In the 1980s, that was probably true. Most of OSI died (X.500, ASN.1 and a few others survived), partly due to its massive scope (like ADA), and partly due to the fact that the authors ignored the IETF and most of the people who implemented the Internet. Much of what OSI tried to do is now being done by the IETF on their own schedule and their own mandate. To the victors go the spoils and the spillage.