Via
http://www.google.com/intl/en/ipv6/ (If you want the pretty graphs, go there)
http://www.youtube.com/watch?v=-Uwjt32NvVA
What if the Internet ran out of room?
In fact, it's already happening.
Vint Cerf, Chief Internet Evangelist at Google, and a founding father of the Internet, discusses the next version of the Internet, IPv6, and why we need it.
Why is the internet running out of room?
Just as phones use a system of phone numbers in order to place calls, every Internet-connected device gets a unique number known as an "IP address" that connects it to the global online network.
The problem is that the current Internet addressing system, IPv4, only has room for about 4 billion addresses -- not nearly enough for the world's people, let alone the devices that are online today and those that will be in the future: computers, phones, TVs, watches, fridges, cars, and so on. More than 4 billion devices already share addresses. As IPv4 runs out of free addresses, everyone will need to share.
How are we making space to grow?
Clearly the internet needs more IP addresses. How many more, exactly? Well, how about 340 trillion trillion trillion (or, 340,000,000,000,000,000,000,000,000,000,000,000,000)? That's how many addresses the internet's new "piping," IPv6, can handle. That's a number big enough to give everyone on Earth their own list of billions of IP addresses. Big enough, in other words, to offer the Internet virtually infinite room to grow, from now into the foreseeable future.
When is the transition happening?
At Google we believe IPv6 is essential to the continued health and growth of the Internet and that by allowing all devices to talk to each other directly, IPv6 enables new innovative services. Replacing the Internet's plumbing will take some time, but the transition has begun. World IPv6 Launch on June 6, 2012, marks the start of a coordinated rollout by major websites and Internet service and equipment providers.
You do not need to do anything to prepare, but if you're interested in learning more and supporting IPv6, check out a few frequently asked questions.
Frequently asked questions
What are IPv4 and IPv6?
IPv4 is the current version of the Internet Protocol, the identification system the Internet uses to send information between devices. This system assigns a series of four numbers (each ranging from 0 to 255) to each device. IPv4 only allows for about 4 billion addresses and the Internet needs more room than that. IPv6 is the new version of the Internet Protocol and expands the number of available addresses to a virtually limitless amount–340 trillion trillion trillion addresses.
Why is IPv6 important?
The Internet is running out of IPv4 addresses. Transitioning to IPv6 enables the Internet to continue to grow and enables new, innovative services to be developed because more devices can connect to the Internet.
What is an IP address?
Just like a phone number helps you communicate with another phone, an IP address (short for Internet Protocol address) is provided to your computer so it can communicate with websites, Internet services, and other devices. IP addresses are numbers that are displayed as strings of letters or numbers, such as 192.0.2.1 (for IPv4) and 2001:db8::1234:ace:6006:1e (for IPv6).
What is World IPv6 Launch?
World IPv6 Launch on June 6, 2012, organized by the Internet Society, is the day participating major websites and Internet Service Providers (ISPs) permanently enable IPv6 and begin the transition from IPv4.
Will I notice anything different? Do I need to do anything to prepare?
You don't need to prepare anything for IPv6; your applications and devices will work just as they did before. This change is to make sure you can keep using the Internet in the future just as you do today.
How long will this transition take?
The complete transition to IPv4 to IPv6 will take time as every website and Internet Service Provider must make the switch. In the meantime, both systems will work together until IPv4 is no longer needed.
Will you be turning off IPv4?
No, IPv4 services will continue to operate as usual.
Do I already have IPv6?
You may be using IPv6 already, visit ipv6test.google.com to find out. Many devices you use already support IPv6; however, the websites you visit and your Internet Service Provider must first enable IPv6 before you can use it.
I don’t have IPv6 yet. How can I get it?
Many major websites and Internet Service Providers now support IPv6, but there are still many more who need to switch. If you'd like to use IPv6, contact your Internet Service Provider asking them to provide you with IPv6 Internet access. You may also need to enable IPv6 on your home router or upgrade to a home router that supports IPv6. For a list of home router manufacturers that support IPv6, start here.
Was there ever an IPv5?
Version 5 was reserved for the Internet Stream Protocol developed prior to IPv6–it was never widely deployed and will not be used publicly.
Posts
I need to start hitting the books again, but I really don't want to.
Doc: That's right, twenty five years into the future. I've always dreamed on seeing the future, looking beyond my years, seeing the progress of mankind. I'll also be able to see who wins the next twenty-five world series.
Yeah, this. It's trivial to remember "local address ending in .50", but "fe80::200:f8ff:fe21:67cf"?
geez, if you do it at home all the time, have a hosts file with shit like "xbox" "living-room-pc" and "gaming-rig" resolve to your tatic, never gonna change iPV6 addresses.
The death of the dynamic IP is fucking sexy. Cannot wait for everyone to get on board.
Let's play Mario Kart or something...
Why don't they make it a trillion trillion trillion trillion trillion trillion now? When will they learn?
Most importantly, does anyone know how this will effect game services like steam and WoW?
Now, most people who are just using their computers for web-browsing won't have a problem with this, ever. They'll just connect, their router will assign them an IP address (192.168.1.105, for instance), and they never need to know how the sausage is made. For more technical users, though, occasionally you have programs that need to go through particular "ports." Your router is essentially a very large wall, filled with doors. The doors all lead to the same room on the other side, and the router points the incoming information in the general direction of your computer. Some information, though, can't handle that, and gets lost. First, need to set up a "static IP," which tells your router "always give this particular device this IP address." Then, you need to set up "port forwarding" so that that information goes to the correct computer, essentially building a hallway from one door to your computer (so it's both impossible to get lost, and impossible for information coming through that doorway to go to any other computer). This is fine if you have variable ports, and can tell computers and servers what port to use, but some older programs (specifically thinking of a lot of video conferencing software, here) don't let you change the port they use. So, when you forward the port, it can only go to the computer you've set up the port forwarding to direct to, even if you have two or more people who'd like to use the software. IPv6 solves this, by getting rid of NAT in the form we currently know it; your computer just gets assigned an IP address that's used by both the router and your ISP to identify itself.
So, this fixes a lot of more technical issues with NAT. On the flip side of that, however, is that a lot of your privacy goes away. Data which would previously be sent to your router, and your ISP wouldn't know what you did with it from there, will now instead be sent to a particular IP address on your network; your ISP will know exactly what device is receiving the information. Is that a big deal? Some of the network guys I know seem to think so. I'm not really sure where I stand on it.
It's just that the router/firewall will be IPv6, and you will have to authenticate with the gatekeeper to initiate any bidirectional chatter with the computer on the other side.
Let's play Mario Kart or something...
Given that IPv6 addresses are user-editable, there's enough plausible deniability if you're accused of doing something you shouldn't.
On top of that, any halfway decent firewall will prevent anybody external from gleaning any useful information about your network.
NAT isn't really a form of security. It's just a thin layer of obfuscation.
All modern OSes have some method for name resolution on an IPv6 network without a DNS server. Mac OSX uses DNS Service Discovery, while Microsoft uses PNRP (Peer Name Resolution Protocol). PNRP is the official replacement, so to speak, for NetBIOS. Both of these protocols are routable (though they may be blocked by default firewall configurations).
What I'm not clear about is what happens if you have a mixed serverless IPv6-only environment.
the "no true scotch man" fallacy.
Thanks for the explanation. I wasn't aware of this at all, and I bemoan any loss of privacy on the net.
the "no true scotch man" fallacy.
I can type one up if people want one.
the "no true scotch man" fallacy.
Why? We can start buying things we actually need under the guise of being IPv6 compatibility and no one who approves purchases is any the wiser.
You're a little confused there. DHCP just allows devices to request an unused IP address on the network. You can do this with both IPv4 and IPv6.
What your describing is Network Address Translation (NAT). There is a network on either side of the router, typically public and private. But the router is only 1 device on the public network, but handles traffic for multiple devices on the private network. For connections from private to public, the router assigns a random port to the outbound connection and then the IP addressing on the packets are rewritten using the router's public IP, instead of the private IP from the original device. So from the public networks perspective, only your router is communicating. The router stores which port goes to which device's connection, so that when the response comes back (to the router's public IP address), it rewrites the IP addressing to the private device's IP and forwards the packets to the private network so the private device receives it correctly. For connections from public to private, the router doesn't know what private device to send the packets to unless you set up port forwarding.
Because you're hiding entire network spaces behind a single IP, NAT lets you have far more devices connected than the address space normally allows. IPv6 has enough IP address space that you don't necessarily have to use NAT anymore, every device can have a public IP.
Love one.
I know internally I could have 50 clients on my network. This was the biggest boom to home networks early on as if I wanted two PCs to get on the internet from my ISP I'd have to pay $10 per machine to get their own IP address. So, suddenly if I wanted all my devices to get on the internet now I'd be paying $60 for internet and a good $100 for IP addresses (because they don't lease in blocks of 10s for $10 like they do commercially).
We're just going to see NAT be rebirthed into IPv6 aren't we?
I'm fine with them tracking what I do online for legal purposes, I'm not part of 4chan and I don't torrent. But, honestly, me paying $200 a month for internet on all my devices is fucking stupid.
will i still be able to type 192.168.1.1 or 10.0.0.1 in my URL bar to configure my router
if the answer is no i'm going to hang myself
In theory the spec says that you would get 2^64 addresses (the subnet block is 64 bits long). In practice, well, who knows what ISPs will end up doing? I still can't get IPv6 support at all.
sometime in the future, no
you probably will have a router that just serves a page like www.routerlogin.net or whatever, though
First thing: you've probably seen an IPv6 address by now. They look like this: 2001:0db8:85a3:08d3:1319:8a2e:0370:7348.
This is analogous to an IPv4 address in that each group of digits represents a subnet, up to the last group, which can either be a subnet or a host.
IPv6 addresses starting with fe80: are local addresses, analogous to 192.168.x.x.
The difference is that you have more subnets, and the subnets are represented in hexadecimal (ie. 00c0 instead of 192).
You also don't need to declare leading zeros. fe80:c0: is the same as fe80:00c0:.
You can skip digit groups in an IPv6 address, if the digits are zero. That's what the double colon ( :: ) means. The address fe80:0000:0000:0000:0000:0000:0000:0001 is the same as fe80::0001. Your OS or client will automatically determine how many digits are missing and fill them in with zeroes. So you could also declare fe80:1234::5678:abcd and your OS will automatically fill in fe80:1234:0000:0000:0000:0000:5678:abcd.
If your address ends with a zero, you can just declare fe80::.
Subnetting is done similarly, by blocking out the bytes that represent a subnet. fe80::/64 gives you a subnet where the first 64 bits declare the network and the bits after (and including) position 65 declare the host.
And remember, this is all in hex, so each group represents 16 bits (unlike IPv4's 8). fe80:1234:5678:abcd::/64 gives you a network where the first four hextets declare your subnet and the last four hextets are open for hosts.
BTW, the loopback address is ::1. That's analogous to 127.0.0.1.
Okay, with me so far? Here's how IPv6 handles automatic allocation of addresses.
When an IPv6 host first appears on a new network, it automatically assigns itself an fe80::/64 address. This is called a tentative address.
It derives its initial tentative address from it's MAC address. MAC addresses aren't strictly unique, and it's always possible that another host has already been given that address by a human being, so if it turns out that there's an address collision, it will randomize part of the address (I don't remember which part) and try again.
The next thing it will do is look for routers by multicasting a nonroutable Neighbor Discovery Protocol (NDP) packet to the local network and waiting for a response. This NDP packet works a little bit like a DHCP discovery packet.
The difference is that NDP is baked into IPv6. You don't need to set up an NDP server separately.
Any device that functions as a router must (as part of the IPv6 specifications) be able to provide an NDP response. The closest router responds to the crying client with an NDP router advertisement packet that includes the subnet hextets and scope and the router address. Routers are also supposed to broadcast next-hops and negotiate routing tables (much like in IPv4) so part of the IPv6 network discovery functionality is to provide the client with a different router address if a better one is available.
Routers also occasionally send out router advertisement packets on their own, which keep routing tables and clients up-to-date.
What about DNS and other DHCP parameters?
Well, that's a tricky part.
The IPv6 standard also supports providing a DNS server list in the router advertisement packet, but that isn't universally implemented yet.
In reality? Most routers that support IPv6 will also have embedded DHCP servers and will just fall back to DHCP when it's necessary to provide DNS server lists to clients.
Edit: fucking smileys.
the "no true scotch man" fallacy.
Most home networks are going to end up on the fe80::/64 local subnet anyway.
Which means your cheap Linksys router is going to be performing NAT just like it always did.
the "no true scotch man" fallacy.
Probably, because IPv4 isn't going anywhere, but in reality, you won't need to configure your router anymore because 99% of the complex shit you had to configure routers to do no longer matters. Set the SSID, security, shove it out of NAT mode and never touch it again.
"for the time being" being when every square inch of the earth has millions of unique IP devices.
I mean, IPv6 won't be outpaced until we cross Kurzweil-level tech wankery.
Let's play Mario Kart or something...
Generally you don't really need to configure your router now, except to set up your WPA2 password or whatever.
A huge ISP (like AT&T) might get a network block of /20. So let's say AT&T gets to use 2600:1000::/20.
They're supposed to offer each and every customer - even their low-end DSL customers - at least a full block of /64.
That gives them 2^44 distinct subnets to work with - enough for 17,592,186,044,416 customers.
Within each of those subnets, the last 64 hextets are assigned to hosts, giving each customer 2^64 hosts to work with.
the "no true scotch man" fallacy.
UPnP works correctly on roughly 1% of mass market routers and you often have to go in an either extend the DHCP lease time to a month or set up static IP configurations for a few systems that need to field incoming connections.
Pft what if AT&T gets 18 trillion customers, WHAT THEN SMART GUY?
This is not true of my parents, nor of most of the people who only use the internet for looking at videos of cute animals, posting vacation photos to Facebook, and emailing their children for tech support.
Or your PPPOE password for DSL.
Right now, the major home router manufacturers still expect you to use the IPv4 address to perform configuration tasks.
I don't know how they plan on making it brainless for end-users to perform config tasks on IPv6. The easiest thing to do would be to give each router a default hostname and tell customers "just type 'linksys' into your browser."
the "no true scotch man" fallacy.
NetGear routers have been silently redirecting "routerlogin.net" to the router's configuration page for years now.
It wouldn't surprise me at all if fe80::1/64 becomes the new 192.168.0.1/24.
the "no true scotch man" fallacy.
Smart!
the "no true scotch man" fallacy.
Yay for archaic equipment.
the "no true scotch man" fallacy.