> BTW, WTF are mebiaddresses?
8 mebi IP numbers, or 8 * 1024 * 1024, see http://en.wikipedia.org/wiki/Binary_prefix
Less than three per cent of IPv4 address space is still to be allocated, after two huge chunks were given to American and European ISPs. ARIN and RIPE, which administer IP addresses on either side of the Atlantic, each received two /8 address blocks in November. A fifth block went to their African equivalent. The moves leave …
> BTW, WTF are mebiaddresses?
8 mebi IP numbers, or 8 * 1024 * 1024, see http://en.wikipedia.org/wiki/Binary_prefix
"Mebi" is a desperate attempt to produce an alternative prefix for 2^20 to distinguish it from "Mega" verbally.
I can't believe anyone is actually using it.
I bet he also says omnibus instead of plain old bus and has a telephone instead of a phone.
...and just use it.
He probably spells "poncey" correctly too........
There are huge chunks of the existing address space which are locked up behind firewalls with no intention of ever letting them route directly. I know that some companies who do have official address blocks in use internally still NAT connections to the outside world.
As others have pointed out the cost of renumber inside a company is considerable. There needs to be some sort of incentive. Why not allow companies with existing large address blocks to rent out their address ranges.
If companies could see a potential profit from re-numbering then they'd be onto it like a shot.
How long do you think it would take HP's management decide they didn't really need to sit on two whole /8s, a shed load of /16s and more /24s than you can shake a stick at if there were $$$ at the end of it.
HP don't own any IP addresses. I doubt very much that they would be allowed to "rent them out". It would open the door for IP squatters doing the same thing that domain squatters do right now.
When IANA is depleted - and I'd guess at February too, the RIRs will (by definition) all have at least one full /8 and a fractional /8. That will last several months for even the busiest (like APNIC, ARIN and RIPE) and AfriNIC and LANIC should last a bit longer than the other three.
Even when the RIRs are out, the LIRs (ie ISPs) will have blocks still to allocate to customers, so you're not likely to phone up an ISP and be told "sorry, we're full" until the back-end of 2012.
Still, what that does is it gives you two years from now to have your IPv6 transition planned, costed, and ready to implement.
There will be people who can only access the internet via IPv6 or 6to4 proxies by 2013 or 2014 at the latest. If you're an AS then you should be aiming for IPv6 by then - 6to4 proxies are going to be a pig.
[Oh, how I wish that the IPv4 space had been allocated a block in IPv6 and routers were required to down-convert packets; the transition would be so much easier]
"...you're not likely to phone up an ISP and be told "sorry, we're full" until the back-end of 2012."
Correct. The final end of the Internet is scheduled for December 21, 2012.
Some companies have bought up other companies and ended up with more than one class A. IIRC Apple has its own + one or two more.
Perhaps Apple has a big plan to use these for iphones etc.
In other news: Doom 4 will be released with only IPv6 support, not IPv4 support.
(hit F3 for grenade launcher?)
... the main question that arises is probably : if I need to access a site that has *only* an IPv6 address, will I be able to do so using my present present browser, router and ADSL line with no changes at my end other than making sure I have the latest operating system and browser updates. And if not, what action will I have to take.
Perhaps somebody at The Reg could either create such a guide, or find a well written one and publish a link to it.
This question shows a lack of understanding of IPv6. An IPv4 device cannot access a IPv6 server.
Think about it: how are you going to express an IPv6 address in an IPv4 packet? Can't be done.
If you have IPv6 support on your PC you could TUNNEL your IPv6 through your ISP's IPv4. But you MUST HAVE IPv6 support with IPv4 tunnelling on your PC.
The answer is that it depends. But given a good NAT implementation by the ISP then you will probably need do do nothing at all. Incidentally a sophisticated NAT implementation by the ISP could probably deal with most of the objections to NAT in other comments - access to internal mail servers for instance - but would require a level of sophistication and interaction between ISP and customer which is almost certainly quite impractical at current support and pricing levels.
to access ipv6 only sites (but hardly any exist right now, and none that are important).
You just run an bit of free software on your machine that creates a tunnel to an ipv6 gateway. Have a look at somewhere like http://www.sixxs.net/faq/ if you're interested.
I played around with ipv6 about a year ago to see what it was all about, but for the end user it turns out to be pretty pointless.
AFAIK the problem for Joe Public is the ADSL router. All mainstream OSes have supported IPV6 for years, as have the routers used by businesses. Even Cisco support IPV6 now. However, try finding an ADSL router that supports IPV6 for less than several hundred £ and you're onto a loser. Having said that, devices like the DrayTek Vigor 120 http://www.draytek.com/user/PdInfoDetail.php?Id=71 are getting there.
There are several routers that are open hardware, on which the manufacturer's firmware can be replaced by Linux (or by a different Linux kernel). Some are designed to be open, some have been cracked by enthusiasts. Linux has supported IPV6 for years. So the cost problem will solve itself just as soon as there is a mass market for IPV6 routers.
> if I need to access a site that has *only* an IPv6 address, will I be able to do so using my present present browser
Yes. For example, to look at ipv6.google.com from a V4-only machine, go to
(The gateway is apparently in Hong Kong, because that's where Google thinks I'm coming from :-)
However, you will never need to do this. No content provider is *ever* going to put content on V6-only, making it invisible to the Internet (unless they don't care about users and customers). If people are prepared to pay $1m+ for a domain name, they'll certainly pay a few thousand dollars for their own IPV4 address, or will sit behind a shared HTTP proxy.
Adding layers, complications, signups, and extra stuff is going to be an annoyance for the end user. Really, it seems like a 6to4 tunnel is a bit of a hack to fudge the two systems. Is it not possible for an ISP to support both IPv4 and IPv6 packets at the same time? Well, it sort-of is...
Tracing route to shake.stacken.kth.se [2001:6b0:1:ea:202:a5ff:fecd:13a6]
from fe80::222:43ff:fe26:78f9%6 over a maximum of 30 hops:
1 * * * Request timed out.
2 * * * Request timed out.
3 * * * Request timed out.
My machine would appear to be IPv6 capable (it can ping itself) but my Livebox does not appear to fully support IPv6, or if it does, Orange themselves don't. I find it odd I can DNS for an IPv6 address, but then not speak to it.
At least I don't have to worry about being shoved behind a NAT. Most Liveboxes over here carry a VoIP phone, and in our case (being dégroupée) it's the ONLY phone, and with a proper local number, not some special-service number. But I'd drop them in a heartbeat (and stuff the contract) if running my mailserver becomes impossible. It's just a damn shame the sharks want €8/month for a fixed IP address (a price that will surely rise to silly figures if addresses "run out").
ipv6 was designed by nerds for nerds and has no appeal to the general net using public whatsoever. And instead of making it truly backwards compatible with ipv4 so that the net could just expand seemlessly, they decided a total rewrite was in order, hence the slow takeup.
Hey, but at least your toaster can have it's own IP address with ipv6!
Not to mention the HUGE security risks of having your PCs directly connected to the internet again, with no NAT to firewall most of the port scanning attacks...! In tomorrow's news: IPv6 enabled windows PCs get hacked on port 445 again!
as per title.
So instead they made a far more efficient version with a lot of address space.
The problem is the dire lack of knowledge about IPv6 and people posting nonsense based solely on rumour and ignorance. IPv6 has a /118 address space for site-local (equivalent to private space in IPv4) and another /118 space for link-local, a new category of private addresses. So you can NAT away to your heart's content.
Face it, when the internet explosion started, fifteen or so years ago, it was a nerd thing. There's now a lot of stuff on top of the basic IPv4 that hides it from the customer. The ADSL router gets its IP address from upstream. The computer gets its IP address from the router. If those machines can handle IPv6 then why should I care as the user?
Yes, there are things that I, specifically, would need to be careful of. I have a hard drive on my network, and occasionally run a server on localhost. But I don't use IP addresses directly.
but it may not be elegant.
The basic plan would be that you designate a single IPv4 address as indicating that the packet is really IPv..( erm 7?) and that extra address information is then found at offset X.
At offset X you have a few bytes to verify that the packet is really the new format plus whatever extra address info is needed.
Machines and routers need software updates to understand the new format but the crucial fact should be that any unupdated router should just pass the packet on as an IPv4 packet that it does not understand. Routing tables would need a bit of clever work to make this work right, but I'm sure it's achievable. Also crucially, old IPv4 packets can continue to be sent and received as before and a machine with an IPv4 address can talk to a machine with a new address without having to have a new address itself.
Well that took me all of fifteen minutes to rough out, I'm sure there are plenty of flaws but nothing insurmountable. Don't tell me that with more time all the brains of the internet couldn't make the general principle workable.
Maybe if IPv6 wasn't such an incomprehensible abortion then you might find more techies happily switching sooner. As it is, why go through that pain until you HAVE to?!
Also, as suggested previously: Intel invent a new 64-bit architecture (IA64) and it bombs. AMD come up with a 32-bit compatible 64-bit architecture (AMD64) and it takes off like a rocket 'cos it's a no-brainer! Maybe the people who put v6 together should've done something similar...
Don't you mean port f120:::2020:3050:0445
No-one's used 126.96.36.199.0 for decades - and when it was used it generally wasn't routable. I doubt if it appears in any BGP tables.
The International Amateur Radio Union (IARU) will undoubtably welcome the cash
brd1#sh ip bgp 188.8.131.52
BGP routing table entry for 184.108.40.206/8, version 24216485
Paths: (2 available, best #1, table Default-IP-Routing-Table)
Advertised to update-groups:
6461 2152 7377
220.127.116.11 from x.x.x.x (18.104.22.168)
Origin IGP, metric 7021, localpref 100, valid, external, best
6320 20500 22822 11164 2152 7377
22.214.171.124 from x.x.x.x (126.96.36.199)
Origin IGP, localpref 100, valid, external
Community: 414208520 414208620
Noooo..... We "hams" are desperate to hang on to our radio frequency assignments, even if some are under used at present. The reason is that often a new technology comes along and without reserved space, there is no place to experiment. I believe things should be similar in the IP world.
Even the IP registries have held back some v4 space for future possible use.... Maybe the 188.8.131.52/8 range should be treated the same way or added to the same pool.
The "first comers" on the Internet (mainly large US corporations) are still using lots of /8 classes as their private (and firewalled) IP space. These addresses are useless (they could be allocated using private subnets) and should be returned.
..."provides many orders of magnitude more address space".
No kidding, that's like saying Elvis liked the odd burger
...that every grain of sand on earth could have an IPv6 address.
if you understand the proper meaning of "orders of magnitude."
Specifically, IPv6 provides exactly 96 binary orders of magnitude, or approximately 28 decimal orders of magnitude, over IPv4. In the common IPv6 notation, which is hexadecimal, IPv6 provides 24 orders of magnitude over IPv4.
Sure, in individual numbers that's a s**tload of space, but in orders of magnitude, "many" covers it quite nicely.
if an order of magnitude was a burger and i ate 96 for lunch then technically/semantically I have eaten many burgers
I use a tunnel for free access to usenet binaries (I love you xnews!). How else am I going to avoid the prying eyes of the copyright cartels? ;)
Arrrrr me hearties!
Many newgroup providers offer an ssl connection option as well. Yes an outsider could know you connect to the newsgroup server but who cares you are simply doing so for work newsgroups :P.
I'm calling it a fake crisis because we already KNEW it was coming and we already have the solution in hand. What's the matter with us that even when we can see that far ahead, we can't do anything about it until we actually run into the wall?
Supports my conclusion that the answer to the Fermi Paradox is that most intelligent species don't survive. Especially given the recent evidence of large numbers of planets, I'm increasingly convinced there must be lots of opportunities for life out there, but the intelligent forms remain hidden. At this point, I'm even quite doubtful we should claim to be an intelligent form. In spite of the lack of evidence, I don't think ALL of them exterminate themselves, but I'm inclined to believe that there are very few survivors--but they are probably watching us and betting quatloos on how long we'll survive. If I was betting, I'd probably bet on a supervirus bioweapon in the next 10 years...
"Universal IM" will happen first. lol
> What I want to know is not "how much space is allocated to top-level
> allocation organizations?" but "how much space is actually being used
> by devices?".
> When the latter maxes out, we have a problem.
Not quite: when the number of IPv4 subnets that are actually being used by devices maxes out, we have a problem. That's much, much sooner than 4 billion public IP addresses actually being used.
A few months ago, the ISPs with opinions were expecting this to happen around 2015 (source: RFC 6036). Which is, like, tomorrow in terms of rolling out v6 everywhere. If you are an ISP and you have no plan, your entire revenue is at risk.
Let me tell you what the majority of ISP's will have as their plan.
But first we need to establish a few facts.
1) ISPs hate P2P because it makes them transmit a tonne of packets.
2) ISPs hate P2P because people use it to download movies and music, which ISP's would very much prefer you to purchase from their very own online stores.
3) ISPs hate VoIP because for the most part, they also are selling you your voice services as well and VoIP hinders their ability to do that quite substantially.
4) ISPs couldn't care less whether you are able to run your own web/mail/game server because they would prefer that you simply use your connection to surf the web and be a good little passive consumer. And hopefully they can figure a way of collecting data about your surfing habits to sell off at a later date at some point.
5) ISPs are greedy and amoral and are constantly on the lookout for new ways of charging more for stuff that they already provide.
So, with that in mind, this is how it will go down;
ISPs will move all their consumer customers to NAT. This of course will break all sorts of things but considering that those things are currently costing them money in the form of running expenses or loss of potential revenue they won't care at all. Most of their ToSs already state that running servers on consumer lines is verboten, and if they don't then these can simply be changed a few months beforehand.
For all the people who complain there will of course be a solution and that will be a "Premium Service" which is nothing more than what we currently have, ie: a proper IP address.
This will come at a price of course.
99% of the proles out there will never even notice that things have changed. Of the few that do, most of them will just shrug and go back to watching Justin Bieber clips on youtube, a few will sign up for the Premium Service and the rest will attempt to churn over to one of the ever shrinking number of ISPs that aren't implementing NAT (yet).
The golden age of the open Internet is nearing its end folks. One day we will all be telling our grand kids about the good old days where anyone could just hook a server up to the 'net and you could even create a private tunnel between your networks! All by yourself! No ISP fees required!
If they ever end up actually rolling out IPv6 then this won't be a problem of course but honestly, have a think about your current ISP. Do you _really_ think they have either the ability or the inclination to make such a move?
I'm not betting on it happening anytime soon. For gods sake we are within _weeks_ of running out of addresses and there is only one ISP in the whole of Australia that is supporting IPv6. The others aren't even _trialling_ it as far as I can tell.
It doesn't help of course about the whole IPv6 lack of backwards compatibility thing. Why they IETF didn't just add a couple more octets to the existing IPv4 is beyond me. I realise they wanted to address some of the perceived flaws in IPv4 but the fact is that the solution they came up with is simply too difficult to implement for todays massive existing networks. If the transition had have been fast tracked back when IPv6 was first introduced in the mid nineties when IPX still ruled the roost in corporate networks then it might have been less of a problem but today IPv4 is so entrenched in both the network infrastructure and staff knowledge sets that it is going to be an unmitigated nightmare.
I might have to consider retiring.
I work for a very large Internet icon that was foundation:fundamental in the growth of the internet. They own a number of B class address spaces that could easily be replaced by NAT + private address space. Like so many other large organizations, they are holding on.
IPv4 address space need a re-org, thats all.
IPv4 barely works for most people anyhow because of NA(P)T. For me it's just there as a legacy technology which I only use to communicate to people who don't have IPv6 yet.
I like to compare IPv4 to ISDN. Back then it was 'the big thing', but today it stales in popularity. Of course I have an ISDN line for increased geekiness, but I don't actually use it much more.
IPv6 will overshadow IPv4 in the same way. IPv6 will enable completely new possibilities like peer to peer web applications or proper VoIP.
Could ICANN just declare that in 12 months time, they will start releasing the private /8 ranges and the root DNSs will start pointing to their new owners? The NAT will still stop access from the outside internet, but machines behind the NAT would not be able to route packets to the outside.
Or am I just showing my ignorance?
well how about looking at the more serious posts here...
Where is my flying car?
I don't get the doomsaying from Anon 16 and others about increased risk...just because machines have their own IPs instead of shared ones doesn't mean it can't be firewalled. Bridging firewalls and 1:1 NAT are both perfectly reasonable solutions and would be easy to implement in ADSL routers.
I work as a network engineer in a UK Uni which has a class b subnet, which is essentially split into class c's for use. Almost every device gets a routable ipv4 address regardless of what it is being used for. Most of the class c subnets we use don't need to be a full class c, its just easier for non-networking people to understand. I suspect that we are not the only University who do this.
If someone turned around and told the University they could sell or lease class c subnets for some serious money with no major drawback, the University almost certainly would, and you could get a lot of IP addresses freed up.