IPv6 is Here

shawn(at)fsu writes "Reuters is running a story that Vinton Cerf of the Internet Corp. for Assigned Names and Numbers (ICANN) says that "IPv6 been added to its root server systems" I like how they said that it will run along side IPv4 for 20 years to get rid of the bugs. A few previous Slashdot stories out of many here, here and here"

Comments

v6 could help solve some net problems

[erick99]

Perhaps we will get to the point where static IP addresses are required. That might help track down spammers and other bad netizens. I'm sure they will find a way around it, but, still, an assigned IP for each user and each piece of hardware would be a good start. And, it would seem that there would be enough to 'round:

Cerf said about two-thirds of the 4.3 billion Internet addresses currently available were used up, adding that IPv6 could magnify capacity by some "25,000 trillion trillion times."

Of course, if v4 runs along side of v6 for 20 years that may mean that it would be harder to implement an IP-per-user scheme. I don't know. But, 20 years should be enough time to work out any bugs:

He said the IPv6 system would run parallel to IPv4 for about 20 years to ensure that any bugs or system errors were weeded out.

Cheers!

Erick

Re:v6 could help solve some net problems

[jerde]

IPv6 is big enough to give a class C subnet to every living person on the planet

Um. IPv6 is big enough to give a Class A subnet to every living person on the planet.

It's big enough to route an entire IPv4 numberspace to every living person on the planet, and to each of their pets, favorite invisible friends, and pieces of furniture.

2^128 is a big, big number.

The point is, they'll be able to "waste" huge swaths of the that numberspace as they build the routing hierarchy, making the network more scalable.

I'm worried about remembering ssh 2031:0000:130F:0000:0000:09C0:876A:130B

:)

- Peter

Re:v6 could help solve some net problems

[Ianoo]

Or, you could just type "ssh pc1.id.isp.com". If IPv6 brings a demand for memorable DNS addresses for everyone, I'm sure the ISPs will provide. Considering you'll most likely be able to put each machine on a static IP within your subsubnet within your ISP's subnet, services like DynDNS will no longer be needed, so you can just register a domain and run your own nameserver.

Re:v6 could help solve some net problems

[spookymonster]

With postal mail, you can send mail anonymously (just don't include your address). When you want to receive mail anonymously, rent a mailbox (either at the post office or at a Mailboxes, etc, for example).

If every IP device gets its own address, but you want to send or receive something anonymously, use a public terminal.

For both snail mail and IP traffic, neither solution is convenient. However, the fact remains that it is still possible.

Re:v6 could help solve some net problems

[dotwaffle]

Don't make me Briticise that... Ok...

number 340282366920938463463374607431768211456
three hundred forty billion trillion.
two hundred eighty-two thousand million trillian.
three hundred sixty-six million trillion.
nine hundred twenty thousand trillion.
nine hundred thirty-eight trillion.
four hundred sixty-three thousand million billion.
four hundred sixty-three million billion.
three hundred seventy-four thousand billion.
six hundred seven billion.
four hundred thirty-one thousand million.
seven hundred sixty-eight million.
two hundred eleven thousand.
four hundred fifty-six.

Now let that be a lesson. Lesser-cousins of ours.

Re:v6 could help solve some net problems

[JPriest]

I got bored one day and figured out there are over a million IPv6 addresses for every square inch of the earth suface (counting water). Also it is not difficult to track somwone down just becasue they have a DHCP assigned IP address. Without DHCP we would have to hire 1 administrator for every person becasue they will have 15 or 20 IP connected devices.

Re:v6 could help solve some net problems

[dgatwood]

If I move from CA to NY, the routers of the world would have to change their tables to be able to get information to me. That is just for one person. Now think of all the people who move or change ISPs.

Easily solved. How about something like this:

• 128 bits of addressing....
• lower 48 bits for the host ID
• next 40 bits for your internet number provider
• top 40 bits for current network routing info

The client could try to send a packet to the static IP number as-is. If it came back with a network unreachable ICMP response, the client would assume that the server moved. Thus, it could use a variation of the existing DNS mechanism to ask who routes for a given host ID part. The root server throws away the top 40 bits and looks at the second 40 bits (the number provider part). It then says "mybigisp.com knows the answer for those 40 bits". You ask dns.mybigisp.com, which might subdivide those 40 bits into a multi-tiered namespace according to where you lived when you joined the network, for example.

Now at that point, you have a regional host in your original home region which says "the last time I heard from myHostID, it was coming from mycurrentisp.com's routing space."

Next, you ask mycurrentisp for the current IP number for myHostID. It does the same multi-tier lookup based on some division of the 48 bit host id part and hands you back the current IP number with a new top 40 bits glued on. Routing then occurs using only the top 40 bits. This still gives you 256 times the current total IPv4 namespace to use for routing, and unlike IPv4, it can be allocated dynamically.

Now from your server's side, your machine thinks it has a static IP. The first router it encounters speaks some IPv6 routing protocol and says "I'm suddenly seeing packets from a device whose number provider ID is (insert 40 bits here). I'll notify my upstream, which eventually would get to your ISP's name server, which would propagate it across to the number provider's name server.

In a scheme like that, you could have a static IP and not have a static IP at the same time. :-) Or maybe I'm just nuts. Hard to say.

Perfect!

[mfh]

FTA: Cerf said about two-thirds of the 4.3 billion Internet addresses currently available were used up, adding that IPv6 could magnify capacity by some "25,000 trillion trillion times."

Perfect for colonization of other planets. If each human being has their own IP, then we would need to pack a whole bunch of planets to require more than that! They aren't kidding when they say they'll run IPv4 with IPv6 for twenty years. In that time, we won't have used even a fraction of a couple percent of available IPs, even if we assign every human being on the planet with one, and every company with a giant block.

Re:Perfect!

[ultrabot]

If each human being has their own IP, then we would need to pack a whole bunch of planets to require more than that!

Binding IP's to humans is arbitrary - it's more realistic to expect that every human with money is going to have several IP addresses (appliances, toasters, whatever), while most of the people in the world will have none.

Also, for some reason, I don't really like the idea of persistent per-human IP addresses. The idea has an Orwellian feel to it.

My IPv6 Rant

[GeorgeH]

I posted How the Internet is broken, how to fix it, and why that's not going to happen, a rant about IPv6 adoption, to my personal site.

Basic idea - include IPv6 over IPv4 tunneling software in Linksys routers. This would allow people to run IPv6 networks in their houses and talk to IPv6 networks elsewhere. This would fix a lot of problems that NAT introduces, and would sidestep the wait for IPv6 ISPs. It would also provide enough of a user base to encourage application developers to include IPv6 support.

Of course, this would kill Linksys' NAT router sales, so they have no incentive to do so, but I like to think it's a good idea.

Re:My IPv6 Rant

[GeorgeH]

It would spur IPv6 adoption, which in turn would make NAT useless once ISPs started providing enough IPs for everyone, instead of their current 1 IP per customer (and sometimes less thanks to DHCP) allotment.

Of course, since public companies are focused on short term sales they might see the IPv6 tunnel sales increase as worth cannibalizing IPv4 NAT router sales.

Re:The bottom line

[Lord Bitman]

yeah, linksys sure would hate for that to happen to its router sales. No company would want to sell four personal firewalls to every home instead of one whole router.
Try again.

ping6 slashdot.org

[caluml]

When will Slashdot get an IPv6 address. Everything supports it - DNS, Apache, etc, nd all they need is to either get an IPv6 tunnel from a broker (the cheap option), or get their ISP to let them have it natively.

Might not be a bad thing.

[Gordonjcp]

After all, registration numbers ended the concept of motor car anonymity, and most people would agree that the advantages far outweigh the disadvantages.

nooo nooo noooooooo!

[mabu]

IPv6 needs to stay in the can until we can figure out a way to solve the spam problem. Right now, RBLs are the most effective method of stopping spam. If IPv6 rolls out, spammers will have exponentially more address space from which to operate and the ensuing spam problem will make what we have now look trivial.

A prerequisite for the rollout of IPv6 must be law enforcement getting off their asses and demonstrating that spammers will get busted for their illegal activities. Otherwise it will take 20+ years to ID and block IPv6 rogue IP space.

you could'a had a v8!

[Anonymous Coward]

aw come on guys! IPV6 has been *DEAD* since the day it was created!

y'know you coulda had a v8! IPV8 that is...

so lets get with the problem and stop using useless junk like ipv6, and start using something that works, seemlessly, like ipv8.

yep, could'a had a v8! www.ipv8.org (obligatory link!)

heh

Re:where are the IPv6 native ISPs?

[schmiddy]

It's the classic chicken-and-egg problem. ISPs know that it will take quite a bit of up-front cash to convert their routers to IPv6. I don't even know to what extent v6 backbone routers exist, but I'd bet their pricey.

Right now.. really the only people that can use v6 are the BSD/Linux folks, as well as (I think) OSX. That's like 5% of the entire Internet desktop users, according to Google's Zeitgeist.

It's a scary thought.. but really, I think critical-mass v6 adoption rests solely on the shoulders of Microsoft at this point. We had better pray that Longhorn comes with it enabled (and that's like 2-3 years from now, at the earliest). If that happens though, it's a sure bet that Linksys et al. as well as lots of ISPs will be on board. I think we'll be waiting a while yet myself, though.

Geographical Allocation

[mindhaze]

I'm sure it's far too late for this now, but what would it take for IPv6 addresses to be assigned geographically? Then, by extension, it'd be amazing if we could just reference an IP Address by the bits that are different. IE: if we're geographically close to the piece of equipment we want to reach, we would only have to reference the last few segments.

That would rock, for sure!

Re:Only if implemented

[redelm]

Sure, you can implement it and twiddle some bits, but it won't be successful if your ISP can't route packets back to you!

It will depend on whether the ISP's router is sufficiently stateful (and lax about MAC) and/or willing to do broadcasts.

This is were a few quiet conversations between security authorities and the larger router manufacturers (Cisco) can have large impacts. "Our routing/auth tables key off MAC for speed and efficiency. We don't do broadcasts to avoid congestion and security concerns." Plausible deniability.