Net shakeup looms as IPv4 resources start running low A new study has called into question the previous received wisdom that IPv4 addresses are running out, forcing the long-heralded move to IPv6. The migration to IPv6 is needed because it offers a massively increased address space as well as advantages in mobility and security. The use of technologies such as Network Address …
There's no technical reason why RFC1918 IPv4 space behind company's firewalls should be an obstacle to taking up IPv6 - Just configure your internal routers to prepend your TLD to the IPv4 addresses and Bobsherunkle*.
*Yes I know this is a bit simplistic, but in principle that's what is needed.
Is it not the case that alot of mobile network providers are assigning private addresses to the end user devices for content access? surely this counters that argument about the upsurge in mobile internet access. Also, did not the numbering authorities issue huge blocks to the colleges & universities - can't these be pulled back.?
"The allocation rate of IPv4 addresses continues to increase due to the growing number of devices that require IP addresses - mobile phones <snip>"
Don't want to be pedantic or owt, like, but isn't IPv6 support a requirement for 3G phones? Or is he saying they need an IPv4 address *as well* as an IPv6 address?
All enlightenment gratefully received. After all, it's been twenty years or so since I did my first "IPv6 is the future, IPv4 is dying" training course. Oddly enough, IPv6 still is the future.
but it looks like a private one. that means that on the outside you could be IPv6. also everyone on you network could be useing just a handful a IPs. still kinda invalidates the who IPgeddon.
(seriously, I thought i was told we would run out by now... and the ozone was going to eat us. is it any wonder so many take reaserchers with a piller of salt? they keep crying "wolf!")
FIRE!
Part of the problem is that big blocks of IP addresses are essentially handed out for free (or a token handling fee) - not as badly as in the early days, when individual universities and companies were handed 17 million addresses each, but far from efficient. GE, IBM, Apple, MIT, Ford and the UK Department for Work and Pensions all hold blocks of 17m addresses each, and HP have TWO such blocks thanks to inheriting DEC's! Do they really need all 33 million addresses? Seriously?
Meanwhile, the university I work for is hogging a whole /16, when a /24 and NAT would be plenty. Once the LEGACY-8 allocations have been revoked and universities bumped to a /24 each, claims we are "running out" might hold a bit of water - but not while so much of the address space is still being squandered so profligately.
Moving to IPv6 still makes a lot of sense, but it's not as if IPv4 space is really in short supply - it's just horribly misallocated at present.
Yes, i agree, just been doing a quick look:
General Electric Company: 3.0.0.0 - 3.255.255.255
Level 3 Communications, Inc, 4.0.0.0 - 4.255.255.255, 8.0.0.0 - 8.255.255.255
IBM Corporation: 9.0.0.0 - 9.255.255.255
Xerox Corporation: 13.0.0.0 - 13.255.255.255
Hewlett-Packard Company: 15.0.0.0 - 15.255.255.255
Couldnt be bothered looking any further. However, those five companies by themselves hold nearly 2.5% of all the IPv4 addresses. (Probably a bit more if you factor in the reserved ones)
"Mismatch between supply and demand in any market creates problems."
No it doesn't. Mismatch creates an opportunity. Mismatch is indication to suppliers to adapt to new situation on market. Is there any real business person that would call high demand "a problem"? Unless of course you run state garanteed monopoly...
So may be hard rule of economics is something like "...creates problems in any scheme where the nature of free market was thought not to apply..."
... because we've been here before. Back before the Internet was the dominant networking technology DEcnet had the same problem - running out of (16-bit) address space as networks became increasingly interconnected - and planned a whole new network-layer infrastructure. The trouble is that you have to migrate your entire network to use the newer protocols before you run out of "old" addresses - otherwise you end up with partial connectivity. DECnet at least had the advantage that the name of the remote system was passed across the application interface so that you didn't have to recode every application to understand the new address format. With IPv6 it's not just every system that has to be upgraded but every application (in principle) before you can start using the increased address space.
It's a chicken and egg situation - not everyone is going to go through an IPv6 migration until they need to; by the time they need to, it's going to be too late.
I think we're going to see a bastardised DNS/network-layer relay that provides backwards compatibility for IPv4 systems wanting to talk to IPv6 networks (the DNS server notes the lack of IPv4 address, allocates a temporary one and relays to the IPv6 network) and that many end systems might continue to use IPv4 as an access protocol for years to come.
It's true that the vast majority of IPv4 addresses are not doing much - they're allocated to various US government bodies and companies that have little need for them. A global reallocation of the address blocks would give everybody some breathing space and cost little.
That said, current NAT strategies, particularly in Asia where the imbalance is greatest, are starting to cause problems of latency and security. But IPv6 rollout requires a great deal more work on the consumer electronics side of things. It's less about the PCs and more about the routers, DSL units and particularly mobile phones.
Is that there was no nice, neat, predefined way to adapt IPv4 addresses on top of it. Yes, you can pu the IPv4 address space inside of IPv6, but they never nicely defined a nice way for a router to "cross over".
If course, there are some who think that 128 bits of address is a bit much, but considering how many things that MIGHT be connected, it may not be. I suspect that they might assign IPv6 addresses to every key on your keyboard (and hav a few left over).
Most (all?) mobile providers run NAT - my iPhone is currently on a 10.x.x.x IP and my 3G dongle with 3 does the same. I agree we need to move to IP6, but factual inaccuracies don't help the cause. Also, there's a *lot* of public address space that's wasted:
http://www.iana.org/assignments/ipv4-address-space/ipv4-address-space.xml
Do IBM et al all *need* those /8s? I think not...
having worked actively on the iEFT IPv6 committees for years, why is IANA not requesting full use of the Class B address space sitting behind so many firewalls/NAT's? I know for firms I've worked for over the last 20+ years that there are numerous companies that begged/B.S.'d their way to Class B's back in the day that will NEVER use all that address space and probably already have 10.'s other non-routable addresses deployed. Having also worked on the DHCP committees the transition can be an orderly one to non-routed external addresses that getting back probably close to 70% of the Class B's would carry the public internet for a while.
As for comments that IPv4 and IPv6 don't overlap well, that's mostly B.S. as the core routers all support both, it's the carp ISP's that don't route that causes most of the trouble as 4's route via 6 networks quite fine thank you very much.
Lazy network admins and carp ISP's are causing most of the problems here. Many ISP's will "sell you a IPv6 address" for an additional fee. Trash the fee and make the addresses available to any customer wanting on, and move all new customers onto IPv6 would help immensely.
>> why is IANA not requesting full use of the Class B address space sitting behind so many firewalls/NAT's?
Because it's not IANA's job. It's not easy to take IP space away from someone once it was allocated to them, no matter how poorly they use it. Though an RIR will stop them getting more IPv4 space until they've made better use of their original allocation. The RIRs have policies which cover this.
Even if /16s (Class Bs in the land of old farts) could be recovered, they're not going to make much difference to the outcome. The world is consuming ~16 Million IPv4 addresses every month. So reclaiming a few hundred /16s would delay the crunch point by a month or two at best. It's not worth the fucking hassle.
Paris icon 'cos she can last longer than a month or two.
having worked actively on the iEFT IPv6 committees for years, why is IANA not requesting full use of the Class B address space sitting behind so many firewalls/NAT's? I know for firms I've worked for over the last 20+ years that there are numerous companies that begged/B.S.'d their way to Class B's back in the day that will NEVER use all that address space and probably already have 10.'s other non-routable addresses deployed. Having also worked on the DHCP committees the transition can be an orderly one to non-routed external addresses that getting back probably close to 70% of the Class B's would carry the public internet for a while.
As for comments that IPv4 and IPv6 don't overlap well, that's mostly B.S. as the core routers all support both, it's the carp ISP's that don't route that causes most of the trouble as 4's route via 6 networks quite fine thank you very much.
Lazy network admins and carp ISP's are causing most of the problems here. Many ISP's will "sell you a IPv6 address" for an additional fee. Trash the fee and make the addresses available to any customer wanting on, and move all new customers onto IPv6 would help immensely.
The comments about the unallocated /8s and the /8s that went to the likes of IBM and HP in the early days are true but irrelevant.
Whingeing about reclaiming inefficiently used /8s of IPv4 is like re-arranging the deckchairs on the Titanic. It's not going to make a difference to the eventual outcome. And even if those /8s could be recovered, it's very unlikely it could be done in time. Can anyone renumber a live global net with SLAs for tens of thousands of nodes in under 12 months?
On average IANA hands out a /8 to one of the RIRs each month. Assuming that rate continues, IANA will have no more space to hand out by August next year. The RIR reserves will go about a year after that. Of course, things could happen much more quickly than that: a last-minute panic to grab for what's left before the shelves are bare. See http://www.potaroo.net/tools/ipv4/index.html
So even if IBM (say) paid for an expensive renumbering to hand back 9/8, that would only give the Internet another month of breathing space. It's just not worth it. BTW I believe IBM and HP use their /8s to provide network services to multinationals, so renumbering these would be even more tricky and expensive. Even if the will was there to do that and it made sense. Which it doesn't.
Wider uptake of IPv6 is the only answer. IPv4 might still be available after the run-out. But it will probably be expensive. And the space might not be usable because ISPs can't be persuaded to route it. Using IPv6 will avoid these uncertainties and other unknowns.
is slightly less terrifying notation for the damn addresses. I mean, 10.0.0.1 any fule can get their head around that with some training.
But try telling someone over the phone to point their browser to "zero eff colon colon bee one colon opening curly brace colon colon ..." (etc) or even remember it yourself for any length of time - not a hope.
Am I just getting old?
This is also my complaint on IPv6. Granted, the numbers are bigger, but even something like 1.2.3.4.5.6.7.8.9.10.11.12.13.14.15.16 looks much prettier than 2001:dead:beef:6666:9999:blahblah. Also, an ugly side-effect of IPv6's implementation is that you actual MAC address will turn up in the IPv6 addy, which means that anyone sniffing your IP has your MAC address, which can be used to bypass MAC filtering. Whoops!
I'm of the impression that there are two main problems with IPv6:
1. Every legacy device that will not ever support it. I don't know about you, but there are certain devices that I am not willing to give up or stop using "just because all they know is IPv4".
That may be something of a moot point, with 6to4 routing and other sorts of solutions to the problem. I guess we'll see.
2. The bigger problem, at least as I see it, is knowing an IP address for a given machine. I have no trouble remembering something like 192.168.1.1 (or what-have-you) but do you really think that I'm going to remember 2001:db8:85a3::8a2e:370:7334 (cribbed from Wikipedia, for those who are wondering)?
Maybe that's not a problem either, given that the IPv6 implementation on the Macintosh (the only place I've seen it used so far) sets itself up automatically....?
For those of you with home routers and a mild-to-moderate sense of adventure, check to see if you can run DD-WRT on your hardware. DD-WRT does support IPv6, although I've never tried to use it. (I mention this only because it might keep a few dozen "unsupported" home routers/access points out of the bin for a while longer.)
Does mobile Internet run over NAT? i.e. when I use my iphone I don't get a public IP allocated from O2 temporarily just for my own handset right (like I do with my home broadband)?
If so that would seem to cancel off the mobile device growth argument as the market can grow massively without any impact on public address space.
My internal network is IPv4. It isn't going to change as I'm pretty sure the router (DLink DIR-100) doesn't do IPv6. But then it is no big deal as the intranet is not connected to the internet. Though, not to mention if I ever put my RiscPC online, there's no IPv6 there... Oh, and my mobile, via Virgin (aka Orange rebranded) had a 10.x.x.x address, which is a much more logical way of providing support for mobile devices.
Anyway. I played with Microsoft's IPv6 networking (XP SP3). I could get a DNS lookup for an IPv6 address, but after that I was not able to do anything. There's a resource for "6bone" online, but these URLs do not seem to be accessible, even trying "telnet6" to get to it directly.
Is there an up to date simple do-this-and-this-will-happen guide to enabling and using IPv6? It would be an interesting exercise to see exactly how much stuff works or breaks.
IP isn't implemented in hardware. Any ISP-supplied modem/router should be capable of doing IPv6, they just need to get off their arse and push a firmware upgrade. My ISP already had to do this once when they went ADSL2+, and this is (in some ways) simpler. *Their* end of the pipe is probably a harder proposition than your end.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
