Editorial

So, what's all this fuss about IPv4/6 anyway?

As every tech news website has been shouting at you for weeks now, IPv4 is coming to the last days of its life, and the world needs to transition to IPv6 to continue to use the internet. But is this true? What’s really going on? Why has IPv4 ‘run out’? And what can you do, right now, to get access to the IPv6 internet? As part of our series on IPv6, we’re going to tackle these questions, and more, to help explain what’s happening, and how you can avoid being left behind.

What is really going on?

IPv4 addresses are allocated, in the first instance, by the ‘Internet Assigned Numbers Authority’ (IANA). The IANA allocates blocks to Regional Internet Registries (RIRs), of which there are five, located around the world: AfriNIC for Africa, ARIN for the US, Canada and parts of the Caribbean, APNIC for Asia, Australia, New Zealand, and neighbouring countries, LACNIC for Latin America and parts of the Caribbean, and RIPE, which oversees Europe, the Middle East, and Central Asia.

The blocks allocated to the RIRs come from the first octet of an IP address, e.g. 209.x.x.x is assigned to ARIN, and it is then ARIN’s responsibility to maintain and subdivide that block to ISPs and companies who request an address range from them, according to whatever policies they decide to establish.

What all the fuss is about is the fact that the IANA is running out of blocks to allocate to the RIRs. Out of 255 possible blocks, at the time of writing, only 7 remain unallocated. It is expected that these will be assigned by the end of the week. Whilst it will be a few more days (or possibly even weeks) before the RIRs fully assign the space to customers, it’s an important wakeup call to people running large networks, that the need to roll out IPv6 is getting more and more important.

RIRs, such as ARIN, have already begun tightening up restrictions surrounding requirements from ISPs who request new address ranges from them. For example, ISPs must be using 80% of the space already assigned to them before being permitted a further allocation. There is also a requirement to begin using the space requested within three months. This is to prevent ISPs from hoarding address space, and possibly auctioning it off to the highest bidder once there are no more ranges left to allocate.

IPv6 gets around this limitation by using a 128-bit integer to represent the address, which makes the number four times as long, and gives many more possible addresses. In fact, the limit with IPv6 is 340282366920938463463374607431768211455 addresses (that's 340 undecillion, 282 decillion, 366 nonillion, 920 octillion, 938 septillion, 463 sextillion, 463 quintillion, 374 quadrillion, 607 trillion, 431 billion, 768 million, 211 thousand and 455). The main difference users will see is that addresses are no longer a string of four numbers, separated by dots, instead, they are composed of hexadecimal values, separated by colons, for example: 2001:0db8:85a3:0000:0000:8a2e:0370:7334. A further part of the IPv6 spec allows for removal of leading 0's, and abbreviations of multiple series' of 0's, so the above address would be expressed as 2001:db8:85a3::8a2e:370:7334, with the double colon indicating where 0's have been removed. This can be used once in an address, to avoid confusion.

Why has IPv4 run out?

IPv4 addresses are, essentially, a 32 bit integer. For example, Neowin has a web server at 209.124.63.215, but expressed as an integer, this is 3514580951 (try loading http://3514580951 and you’ll see a reply from our servers). Because of this, the maximum number of possible IP addresses, is limited to 232-1, aka 4,294,967,295 addresses.

Not all of these addresses are in use on the internet, but there are various reasons why they cannot be used. For example, certain ranges like 10.x.x.x and 192.168.x.x are reserved for use on private networks (most of you will have seen one of these if you have a NAT router at home). There are other ranges reserved for other purposes, but there are too many to list fully here.

Another limitation is caused by the way that routing works on the Internet. In order to keep routing tables (essentially a map that tells routers where to send packets, based on the destination) to a manageable size, the concept of subnets was created. Subnets enable address ranges to be grouped together, with a single destination. Neowin exists within the 209.124.63.x subnet, and any traffic directed to an address within that subnet is sent to a router at Stardock’s datacenter, which then passes it to the relevant server. The downside to this is while the subnet allows for 254 machines to be active within it, we don’t actually make use of that ability, and the same is true for a large number of networks on the internet. Most are used very sparsely, but due to routing protocols, there is no way to allow machines on other networks to ‘fill the gaps’.

Is there anything else ISPs can do to delay the switch?

One thing that is starting to be seen, particularly with mobile operators, is implementation of carrier-grade NAT. This, essentially, is the same as the technology built into your home router, but on a far larger scale. This technology doesn't scale very well though, and doesn't allow for public access to ports on the consumer device, which will be off-putting to many. For these reasons, it's unlikely that CGN will be deployed on home-to-ISP connections, and is likely to remain a technology only implemented in the mobile arena where the number of simultaneous connections is low, and there is no need for publicly accessible ports.

How can I access the IPv6 Internet?

Here, you have a number of options, as transition technologies have been under development for quite some time now. But firstly, it's probably worth giving your ISP a call, and finding out when they plan to roll out IPv6 to consumers, and if there is a trial you could sign up for. If you're lucky, they may have already launched their IPv6 service.

If you're out of luck there, then you have a number of options at your disposal. Teredo is an IPv6 tunneling technology that has been built into Windows since XP SP2, with clients also available for Mac and Linux. Teredo encapsulates IPv6 traffic inside UDP packets, which allows them to transition through most NAT firewalls with no additional configuration. The technology uses a set of servers to direct your traffic to a tunnel relay near to your destination, and relays, which have both IPv4 and IPv6 technology, to forward your traffic to IPv6 hosts. The address prefix of 2001::/32 has been reserved for Teredo communications.

Possibly the easiest option to setup though, is connectivity via a tunnel broker, such as the one provided by he.net. These can be configured on Windows with just four commands, and you're ready to go. Again, like Teredo, this technology relies on passing your communications via a host that is enabled for both IPv4 and IPv6. However, this method may not be supported by all routers.

So, is it worth setting up an IPv6 connection?

The answer for most people will simply be no. IPv4 isn't going to be phased out any time soon, and as the majority of the world doesn't yet have IPv6 access, websites will ensure that they continue to provide IPv4 connectivity to customers through one method or another. If you wish to score geek points with your friends, then by all means, setup a tunnel and enjoy, but the average consumer will run into no issues waiting for their ISP to roll out native IPv6 connectivity.

Image Credit: ThinkBroadband

Report a problem with article
Previous Story

Amazon password flaw

Next Story

Google brings 'Instant Previews' to the iPhone

54 Comments

View more comments

No, there is no need to 'redirect protocol 41'. Protocol != port.
I can launch the aiccu client on any of my machines (1 only at any given time mind you) and it connects just fine. even when I've moved to another network.

cybertimber2008 said,
No, there is no need to 'redirect protocol 41'. Protocol != port.
I can launch the aiccu client on any of my machines (1 only at any given time mind you) and it connects just fine. even when I've moved to another network.

That's because it's using AYIYA, and not true 6in4. It's encapsulating the IPv6 packets inside the payload of either UDP or TCP. Which just adds more overhead (but fixes the problems when behind a NAT). So it's not protocol 41 that comes inside the IPv4 payload, but TCP or UDP.

https://www.sixxs.net/tools/ayiya/

True 6in4 from a computer behind a NAT will still require you to forward protocol 41 to one PC. Some routers don't let you single out a protocol (other than TCP and/or UDP) and will require you to DMZ the entire thing.

IANA have asked RIR's to start hunting down companies that have excessively large blocks allocated and are not in use. The local University near me has been asked to return the large bulk of its IPs as they just moved them all to NAT internal addressing instead of one publicly routable IP per lab.

But again this will only help delay it by a few months. Regardless, when we do run out of allocations consumers are not likely to see problems for quite a while.

Sylar2010 said,
What i still dont get is there any end user effects of this?

Once RIRs assign all their /24's to clients, some organisations return their blocks to the pool and some other last-time measures to delay the inevitable -
- new servers at the datacentres will have to be allocated in IPv6 only space.
- their share will take off the current ~0% and will start to grow.

Clients who are not ready for IPv6 will not be able to view the servers directly.
There will probably be services that will offer ability to connect to a site which will fetch the IPv6 only content for a IPv4-only client.

That will not last long.

Then eventually end users and ISPs will start to finally move somewhere. Install patches, enable IPv6 on their ancient XPs.

Carriers and advanced people are IPv6 ready for now.

The good news is that APNIC clients (China, Japan, S.Korea) will be hit first by exhaustion of IPv4 space, so we in Europe and America will be able to see how the Asian peoples are faring, and what will be their pitfalls, so we don't repeat same errors.

Carrier grade NAT devices (where normal ISP clients would be assigned private addresses) is not gonna work. Just several P2P clients would bring such device to a screeching halt.

So don't try to delay the inevitable. Use time you have to study IPv6 and make yourself dual stack: IPv4/v6 nodes. Deploy the infrastructures in your companies. It's cool. The more brave and technically advanced people will start working on this, the better it will be for the rest of the people.

Sylar2010 said,
What i still dont get is there any end user effects of this?
Has your computer ever told you it had trouble getting an IP address, so you can't access anything on your network? Well, that, except you can't access certain (or any) online material.

EastExpert said,
The good news is that APNIC clients (China, Japan, S.Korea) will be hit first by exhaustion of IPv4 space, so we in Europe and America will be able to see how the Asian peoples are faring, and what will be their pitfalls, so we don't repeat same errors.

I'm sure you meant no harm, but a bit more discretion with your words would have been better.

EastExpert said,

Once RIRs assign all their /24's to clients, some organisations return their blocks to the pool and some other last-time measures to delay the inevitable -
- new servers at the datacentres will have to be allocated in IPv6 only space.
- their share will take off the current ~0% and will start to grow.

Clients who are not ready for IPv6 will not be able to view the servers directly.
There will probably be services that will offer ability to connect to a site which will fetch the IPv6 only content for a IPv4-only client.

That will not last long.

Then eventually end users and ISPs will start to finally move somewhere. Install patches, enable IPv6 on their ancient XPs.

Carriers and advanced people are IPv6 ready for now.

The good news is that APNIC clients (China, Japan, S.Korea) will be hit first by exhaustion of IPv4 space, so we in Europe and America will be able to see how the Asian peoples are faring, and what will be their pitfalls, so we don't repeat same errors.

Carrier grade NAT devices (where normal ISP clients would be assigned private addresses) is not gonna work. Just several P2P clients would bring such device to a screeching halt.

So don't try to delay the inevitable. Use time you have to study IPv6 and make yourself dual stack: IPv4/v6 nodes. Deploy the infrastructures in your companies. It's cool. The more brave and technically advanced people will start working on this, the better it will be for the rest of the people.

XP natively supports IPv6 as of SP2 (this is in addition to Teredo). Vista and 7 (along with Server 2003 R2 and later) are dual-stack out-of-the-box. Windows 2000 (Professional and Server) require SP4; however, even that antique OS can do IPv6 natively. As long as you have a router with third-party firmware and have it set up properly (I used AICCU and a SIXXS tunnel combined with either DD-WRT or OpenWRT/X-WRT for initial IPv6 LAN testing) I could serve IPv6 LAN-wide, wired OR wireless.

Sadly, none of ANY of this actually matters under any conditions. The simple truth is that IPv6 is already dead before it has even started. Here are a few simple truths:

1. Your ISP isn't ready for it. In fact, its probably not even close. Chances are not only would the ISP have to update their hardware and network, they would ALSO have to update the hardware of ALL its customers. Is Tuesday good for you, because that is when everyone else is staying home to have their hardware updated. Your window is between then and 4 years from now. YOu have have a few hundred dollars ready to cover the costs for them, right?

3. Your router isn't ready for it either. But that's easy to fix. Just pop out and buy a new router. Oh wait, that's going to be a problem.

4. Your new router supplier isn't ready either. Not even if you get a third party firmware. Not that it would matter, since the hardware your ISP gave you has no idea how to handle IPv6. Good news - you just saved the cost of a new router!

5. Your favorite software isn't ready, and chances are NEVER WILL BE. Unfortunately the people who wrote IPv4 really didn't think it would ever need to grow. As such, the IPv4 libraries absolutely do not support address parsing. Each and every app/library does that itself, and thanks to the IPv4 standard it is rigidly bound to a 32 bit address with no room for flexibility. Chances are your favorite software is either out of date or written by a company or individual who is no longer available. Sorry. At least there is a chance your browser supports it, and hey - as long as the only way you use the internet is via the brower you are fine. If, on the other hand, you play an online game, use a game console, IRC, etc chances are you'll be looking to pay for an alternative if it exists. Better hope that doesn't mean mission critical software for your business huh?

So, IPv4 was written to be all we'd ever have and all we'd ever need. Software developers had no choice to follow and as such. Most network hardware follows suit- to keep it simple and stable AND keep costs down to maximize profile. Your ISP built their network on that hardware.

It's great to think that IPv6 is the solution, but no one EVER asks how much it will cost or how long it will take to update all that hardware and software.

Buddy can you spare a 100 trillion dollars? And again, everyone is staying home tuesday to wait for your ISP repair guy right?

1. Most ISPs have had IPv6 in testing for a long time, and are looking to roll it out in the coming months (see my article about IPv6 readiness from ISPs)

3?) A lot of routers will be able to support it via a firmware upgrade

4) New routers will definitely be getting IPv6 support, either out of the box, or via a firmware upgrade, if that wasn't the case, router manufacturers would lose a lot of money to the competition

5) This is simply untrue. Most software these days works fine with IPv6. If I disable IPv4 on my network here (I have native IPv6 through some complex method), then I've found only one piece of software that stops functioning, and it's not really critical.

I don't think there is an impending need to migrate to IPv6, otherwise the situation right now would be different. IPv4 addresses and NAT can still take care of those situations in which IPv6 is just not possible.

In an ideal world, ISP routers would fetch both an IPv4 address with NAT, and an IPv6 subnet, and the IPv6 ready OSs would autoconfigure their IPs on both networks. I think that's the most we can expect for the years to come. At one point in some decades maybe we'll be able to ditch IPv4 altogether, but dual stack seems to be the way to go.

DaveLegg said,
1. Most ISPs have had IPv6 in testing for a long time, and are looking to roll it out in the coming months (see my article about IPv6 readiness from ISPs)

3?) A lot of routers will be able to support it via a firmware upgrade

4) New routers will definitely be getting IPv6 support, either out of the box, or via a firmware upgrade, if that wasn't the case, router manufacturers would lose a lot of money to the competition

5) This is simply untrue. Most software these days works fine with IPv6. If I disable IPv4 on my network here (I have native IPv6 through some complex method), then I've found only one piece of software that stops functioning, and it's not really critical.

1. True; Comcast and Verizon (the two largest broadband ISPs in the US) are rolling out IPv6 now in national trials. Verizon is also a Tier 1 ISP for other ISPs (largely via Verizon Business, the former Worldcom and UUnet), which means that a lot of the smaller ISPs will also be dragged in (albeit kicking and screaming in some cases).

3. Most home routers can indeed support IPv6, either directly or via third-party firmware; this is true of most Linksys routers (especially the WRT54G and clones) using third-party firmware.

4. New routers are all pretty much including IPv6 support directly these days (if not in their own firmware, then by supporting third-party crossgrades, such as DD-WRT or OpenWRT/X-WRT). Even Netgear has changed their tune, and added IPv6 and third-party firmware options with their own current-model routers.

5. Every OS (commercial or open-source) supports IPv6 natively today with few exceptions (none of which are on the desktop or in the server closet; Microsoft added IPv6 to XP with Service Pack 2 and to Windows 2000 with Service Pack 4, while Vista/7/Server 2003 R2 and later shipped with IPv6 out of the box); OS X has supported IPv6 since 10.5 (Leopard); Linux distributions with the 2.6 kernels have it at their core, and it can be back-patched into 2.4 kernels.

A few comments on what I've read in the comments.

So far this month, 35.6 million IP addresses have been given by RIRs to ISPs. So giving back addresses is only a short fix and takes time to occur. Block 45 was given back and is still not flagged as available for use by ARIN.

IANA running out of addresses is just phase 1 of the depletion mechanism. RIRs still have addresses to give ISPs, but once they have run out, there will be none left from IANA.

So my ISP has plenty of spare IPv4 addresses. Great, but the Internet is a global system. If you are trying to access a resource (website, telnet, P2P etc) on the internet that is IPv6 and your ISP is only doing IPv4, then you will not be able to access that resource. "But you can just use 6over4" ... I could, but I couldn't ask my parents to do that, they would give me a blank expression and roll their eyes!

Some ISPs say that there is no demand. There is demand, but they don't do anything about it. Do they really have a logging system recording how many people of asked? Of course not, they just fob people off.

I think ISPs need to be a bit more open. Yeah right, thats never going to happen. If they said they wouldn't be able to do IPv6 for 2 years, some pepole will just move ISP. There was an earlier editorial on Neowin that asked ISP's their status, but hardly any responded. This was one of the first media reports I've seen recently that did that. A wonderful start. Lets hope the media moves from just reporting the issue, to asking more searching questions of the ISPs.

mikebiacsi said,
Is Neowin.net IPv6 ready??

Unfortunately not, at the moment it's out of our hands as IPv6 connectivity just isn't available to us at this time. Rest assured though, I keep bugging the relevant people to try to make it happen.

Ridgeburner said,
I have enough trouble remembering my IP address as it is, now I need to know a number 4 times as long??? ugh.

Meh. That's what DNS is for. chances are IPv6 addresses are gonna be almost fixed anyway. Just get a dns name for it

Commenting is disabled on this article.