Forgot your password?
typodupeerror
The Internet Networking

Comcast Plans IPv6 Trials In 2010 173

Posted by timothy
from the wait-for-the-cross-examination dept.
Mortimer.CA writes "In a weblog posting, Jason Livingood, Executive Director of Comcast's Internet Systems has stated that they're beginning public trials of IPv6; Comcast hopes 'that these trials will encourage other stakeholders to make plans to continue, or to begin, work on IPv6 in 2010 so that all stakeholders do their part in ensuring the future of the Internet is as bright and innovative as it has been in the past.' Interested guinea pigs can volunteer at Comcast6.net (FAQ). Those who have IPv6 connectivity via other means can check out their IPv6-only web presence."
This discussion has been archived. No new comments can be posted.

Comcast Plans IPv6 Trials In 2010

Comments Filter:
  • by BitZtream (692029) on Wednesday January 27, 2010 @05:57PM (#30925604)

    I have no ipv6 at this location and it loads just fine here, not exactly 'ipv6 only' like the Dancing Kame ...

  • IPv6? (Score:5, Funny)

    by Delwin (599872) * on Wednesday January 27, 2010 @05:58PM (#30925620)
    Pinging ipv6.comcast.net [68.87.64.59]

    woops.
    • Re:IPv6? (Score:5, Informative)

      by mmontour (2208) <mail@mmontour.net> on Wednesday January 27, 2010 @06:28PM (#30926218)

      Pinging ipv6.comcast.net [68.87.64.59]

      It works for me.

      $ ping6 ipv6.comcast.net
      PING6(56=40+8+8 bytes) 2002:1159:44ef::226:48ff:fe12:a9a7 --> 2001:558:1002:5:68:87:64:59
      16 bytes from 2001:558:1002:5:68:87:64:59, icmp_seq=0 hlim=52 time=235.216 ms
      16 bytes from 2001:558:1002:5:68:87:64:59, icmp_seq=1 hlim=52 time=245.426 ms

      This is through an Apple airport base station via whatever tunnel provider it uses for its IPv6 support. No manual setup, just click the buttons to turn IPv6 on and to block incoming connections.

      • Re:IPv6? (Score:3, Informative)

        by Marauder2 (82448) on Wednesday January 27, 2010 @06:56PM (#30926784)

        Looks like ipv6.comcast.net has both A (IPv4) and AAAA (IPv6) records.

        $ host ipv6.comcast.net
        ipv6.comcast.net has address 69.252.76.96
        ipv6.comcast.net has address 68.87.64.59
        ipv6.comcast.net has IPv6 address 2001:558:1002:5:68:87:64:59
        ipv6.comcast.net has IPv6 address 2001:558:1004:9:69:252:76:96

      • Re:IPv6? (Score:3, Interesting)

        by fluffy99 (870997) on Wednesday January 27, 2010 @11:24PM (#30929226)

        Pinging ipv6.comcast.net [68.87.64.59]

        It works for me.

        $ ping6 ipv6.comcast.net
        PING6(56=40+8+8 bytes) 2002:1159:44ef::226:48ff:fe12:a9a7 --> 2001:558:1002:5:68:87:64:59
        16 bytes from 2001:558:1002:5:68:87:64:59, icmp_seq=0 hlim=52 time=235.216 ms
        16 bytes from 2001:558:1002:5:68:87:64:59, icmp_seq=1 hlim=52 time=245.426 ms

        This is through an Apple airport base station via whatever tunnel provider it uses for its IPv6 support. No manual setup, just click the buttons to turn IPv6 on and to block incoming connections.

        The 2002 prefix on your ipv6 address says you're using 6to4 address translation/tunneling. The ipv4 address at the time was 17.89.68.239. I'm not sure if its your computer doing the 6to4 tunneling or your airport. I'm thinking it's the computer as its using the 2002 address as opposed to the router doing it all in the background.

        • by mmontour (2208) <mail@mmontour.net> on Thursday January 28, 2010 @12:20AM (#30929522)

          I'm not sure if its your computer doing the 6to4 tunneling or your airport. I'm thinking it's the computer as its using the 2002 address as opposed to the router doing it all in the background.

          The tunnel is established by the airport; I'm not running any 6to4 stuff on the LAN computers (Macbook, couple of Linux boxes, and an OpenBSD instance running in a KVM virtual machine). They just auto-configure themselves on the /64 announced by the router. The LAN computers can 'ping6' each other as well as external sites like ipv6.google.com.

    • by harmonise (1484057) on Wednesday January 27, 2010 @07:35PM (#30927336)

      woops.

      I know! What where they thinking by letting people using IPv4 also see that content?

  • by creimer (824291) on Wednesday January 27, 2010 @05:58PM (#30925632) Homepage
    Nope. Can't see the IPv6-only web presence from my IPv4-only internet. I guess it got slashdotted.
    • Re:Oh, well... (Score:4, Informative)

      by jbb999 (758019) on Wednesday January 27, 2010 @06:04PM (#30925756)
      I can see it on my ipv6 connection, it's on 2001:558:1002:5:68:87:64:59 and seems to work :) For those on the UK wanting an ADSL ISP with ipv6 support I recommend Andrews & Arnold (http://www.aaisp.net.uk/) who have been doing this for years now and provide native or tunneled ipv6 and full ipv6 static addresses to their customers on request. Just a happy customer of theirs :)
      • by gbjbaanb (229885) on Thursday January 28, 2010 @06:05AM (#30931244)

        More to the point, which ADSL modem/router are you using that supports IPv6, 'cos it aint no use if my PC and my ISP support it if there's a big block in between the 2 :(

  • by denis-The-menace (471988) on Wednesday January 27, 2010 @06:01PM (#30925692)

    I know most IP6 fan will say that you don't need them but you just know when the smoke clears Joe customer will still get ONE Address.

    Besides, most IP-enabled toys wont like IP6 (Wii, VOIP boxes, etc.)

    • by Manip (656104) on Wednesday January 27, 2010 @06:06PM (#30925796)

      It is very hard to block NATs even if they aren't allowed.

    • by BitZtream (692029) on Wednesday January 27, 2010 @06:06PM (#30925814)

      Its pretty hard to stop someone from using a NAT. Comcast can't really tell the difference between a NAT and a single machine without deep packet inspection.

      At which point you just sue them for invasion of privacy, not that you'll get anywhere but its a neat idea.

      The other side to that is that your IPv6 router can deal with helping IPv4 devices communicate over the IPv6 backbone as long as the backbone does the proper bridging (according to the protocol) back to IPv4, which they'd surely have to if they don't intend to break of the Internet and become their own useless island.

      In short, some very smart people already thought of that problem when designing the system.

    • by tepples (727027) <{tepples} {at} {gmail.com}> on Wednesday January 27, 2010 @06:08PM (#30925858) Homepage Journal

      you just know when the smoke clears Joe customer will still get ONE Address.

      As I understand it, the best practice is for an IPv6 ISP to give out a /64. That's still relatively one four-billionth of the space they're giving out now.

      • by amorsen (7485) <benny+slashdot@amorsen.dk> on Wednesday January 27, 2010 @06:17PM (#30926020)

        No, that is not allowed (well the police won't stop them, but it's definitely not best practice). Best practice was originally a /48, but now ISP's are allowed to cut all the way down to a /56 if they feel a /48 is too much.

        You shouldn't put hosts in anything but a /64, and some don't think there should exist non-/64 unicast networks at all. Personally I believe that at least /128 should be allowed.

        • by dotwaffle (610149) <slashdot&walster,org> on Wednesday January 27, 2010 @07:07PM (#30926966) Homepage

          Let's say your ISP has a /32. The ISP uses a /64 for every point-to-point link between their router and your home router, and you have a /64 within your own home. Additionally, you have a second /64 reserved for you to make VoIP easier. Then, your ISP can clearly only have 1.1 billion customers.

          I realise the above is a bit silly, but seriously, there are enough /64s for everyone. There is no need for a /128, no need for a /126, no need for anything but a /64.

          Even if the ISP was "wasteful" and allocated each residential customer a /56 to do whatever they want with, their /32 will be able to support 16.7 million customers. If you've got more than 16.7 million customers, you just get another /32, in 2000/3 there are 500 million /32s.

          I'm waffling. /64 is fine.

          • by petermgreen (876956) <plugwashNO@SPAMp10link.net> on Wednesday January 27, 2010 @07:49PM (#30927474) Homepage

            I realise the above is a bit silly, but seriously, there are enough /64s for everyone. There is no need for a /128, no need for a /126, no need for anything but a /64.
            The trouble is the ipv6 autoconfiguration mechanisms were designed arround giving each subnet a /64 so if you only have a /64 you either have to limit yourself to one subnet (e.g. no seperate subnet for a segregated wifi network) or configure all your machines manually (and in the case of XP configure them from the command line!)

            there are enough /56's for everyone, hell there are enough /48s for everyone (personally I preffered the old reccomendation of giving every site a /48 because it made it clear where the site number ended and the subnets within the site began).

            Unfortunately you can bet crappier ISPs will give a /64 at best and a /128 at worst.

        • by fluffy99 (870997) on Wednesday January 27, 2010 @11:45PM (#30929350)

          No, that is not allowed (well the police won't stop them, but it's definitely not best practice). Best practice was originally a /48, but now ISP's are allowed to cut all the way down to a /56 if they feel a /48 is too much.

          You shouldn't put hosts in anything but a /64, and some don't think there should exist non-/64 unicast networks at all. Personally I believe that at least /128 should be allowed.

          The first 64-bits are the "network" portion of the address, and the second 64-bit chunk is the interface portion (ie the ipv6 version of your mac address). I'm ignoring multicast for the present. For normal unicast, you can't subnet smaller than a /64. If your ISP is following the standard, they can't give you bigger than a /48 for your site.

          It's also a bit of a myth that ipv6 allows for 2^128 addresses. That's not really true given the first several bits define the address type, not all of the TLAs have been assigned, some of the prefixes are special (like 6to4, and terado), 64-bit host id's uniqueness (generally derived from the 48-bit mac address), ranges set aside for multicast, link-local, non-routable addresses, etc.

          Still ipv6 is a massive expansion of the available range, and solves many routing difficulties. It's also much more complicated and has some drawbacks.

          • by amorsen (7485) <benny+slashdot@amorsen.dk> on Thursday January 28, 2010 @08:47AM (#30931968)

            The first 64-bits are the "network" portion of the address, and the second 64-bit chunk is the interface portion (ie the ipv6 version of your mac address). I'm ignoring multicast for the present. For normal unicast, you can't subnet smaller than a /64.

            It may not be allowed, but it is widely deployed. Not with hosts in those subnets, but it is fairly popular with router-only subnets.

            If your ISP is following the standard, they can't give you bigger than a /48 for your site.

            If you can demonstrate need, you can get up to a /32 even as a non-ISP. Obviously demonstrating the need for such a large allocation is a bit theoretical.

        • by JSBiff (87824) on Thursday January 28, 2010 @02:46AM (#30930252) Journal

          Yes, I know the IPv6 address space is galactically huge, but what exactly good purpose is served by giving each customer 1.8*10^19 addresses? Seems a bit excessive, doesn't it? Wouldn't most customers be fine with 16 bits of host/subnet (obviously, there might be som), and the rest of them shouldn't conceivably need more than 32 bits of their own address space? (And if someone needs/wants more than 32-bits of addressing assigned to them, then, sure, by all means, give them 48 bits). But why, 'by default', give people so many addresses I don't even know the name of numbers that large? (18 quintillion, I guess?)

    • by Cato (8296) on Wednesday January 27, 2010 @07:16PM (#30927078)

      As the website explains, one of Comcast's 3 transition strategies is based on DS-Lite, which essentially means a big provider-based NAT that allows IPv4 only devices such as games consoles to connect via a new IPv4/IPv6 home router (dual stack) over v6 infrastructure to an end server that is v4 based.

    • by Fred_A (10934) <fred AT fredshome DOT org> on Thursday January 28, 2010 @09:14AM (#30932174) Homepage

      I know most IP6 fan will say that you don't need them but you just know when the smoke clears Joe customer will still get ONE Address.

      My ISP gives me (or anyone who cares enough to activate the free option) a /64 IPv6 subnet. It wouldn't make much sense if it didn't.

      I'll grant you that it'll be a while before the various gadgets (or even the software) play nice with IPv6.

  • by Joe The Dragon (967727) on Wednesday January 27, 2010 @06:03PM (#30925734)

    what is the per ip cost? $5? WILL there cable boxes also start useing ipv6? they use ipv4 now.

  • by skogs (628589) on Wednesday January 27, 2010 @06:04PM (#30925762) Journal

    I've been waiting for mediacom to roll out some DOCSYS 3 / IPv6 forever. This little town I happen to be in, has excellent infrastructure and is physically capable of running it -- unlike most cities. This town is dependent only on major hardware upgrades, not cable plant upgrades.

  • by tepples (727027) <{tepples} {at} {gmail.com}> on Wednesday January 27, 2010 @06:06PM (#30925798) Homepage Journal
    The main page mentions tunneling IPv4 over what it calls "Dual-Stack Lite technology (aka DS-Lite)". But Comcast must not have been aware of Nintendo's prior use of "DS Lite" for a handheld video game system with Wi-Fi support. Do Nintendo video game consoles even support IPv6?
  • IPv6 only test... (Score:5, Informative)

    by nweaver (113078) on Wednesday January 27, 2010 @06:06PM (#30925804) Homepage

    ipv6.google.com [google.com] is IPv6 only, and if you can reach it, you are IPv6 enabled.

    We actually used this for the IPv6 test in Netalyzr [berkeley.edu] as the basis of the IPv6 connectivity test. Our servers don't have IPv6, but we have a small amount of javascript on the analysis page that tries to fetch the logo from IPv6.google.com and reports success or failure back to the server.

    • by belphegore (66832) on Wednesday January 27, 2010 @06:22PM (#30926106) Homepage
      ...except that it's not:

      [craig@Puck:~]$ host ipv6.google.com
      ipv6.google.com is an alias for ipv6.l.google.com.
      ipv6.l.google.com has address 208.67.219.132
      ipv6.l.google.com has IPv6 address 2001:4860:800b::69
      ipv6.l.google.com has IPv6 address 2001:4860:800b::68
      ipv6.l.google.com has IPv6 address 2001:4860:800b::63
      ipv6.l.google.com has IPv6 address 2001:4860:800b::6a
      ipv6.l.google.com has IPv6 address 2001:4860:800b::93
      ipv6.l.google.com has IPv6 address 2001:4860:800b::67

      • Re:IPv6 only test... (Score:3, Informative)

        by molo (94384) on Wednesday January 27, 2010 @06:36PM (#30926382) Journal

        208.67.219.132 is OpenDNS.

        -molo

        • by belphegore (66832) on Wednesday January 27, 2010 @06:52PM (#30926682) Homepage
          Ok. So it's only ipv6 if your DNS provider doesn't return IPv4 records for it... It's still not a good test for IPv6 connectivity. A better test for IPv6 connectivity would be, you know, sending an IPv6 packet and seeing if it gets through.
          • by RoFLKOPTr (1294290) on Wednesday January 27, 2010 @07:39PM (#30927372)

            Ok. So it's only ipv6 if your DNS provider doesn't return IPv4 records for it... It's still not a good test for IPv6 connectivity.

            Yes it is. A good DNS provider won't return records when there are none. OpenDNS earns money from ad placement on their bad hostname page, so when there isn't a valid record to a hostname, they return a server of their own. An honest DNS provider is a great test for IPv6 connectivity, though.

      • by mmontour (2208) <mail@mmontour.net> on Wednesday January 27, 2010 @06:37PM (#30926388)

        ipv6.l.google.com has address 208.67.219.132

        Not from here:

        $ host ipv6.google.com
        ipv6.google.com is an alias for ipv6.l.google.com.
        ipv6.l.google.com has IPv6 address 2001:4860:800b::93
        ipv6.l.google.com has IPv6 address 2001:4860:800b::63
        ipv6.l.google.com has IPv6 address 2001:4860:800b::67
        ipv6.l.google.com has IPv6 address 2001:4860:800b::68
        ipv6.l.google.com has IPv6 address 2001:4860:800b::69
        ipv6.l.google.com has IPv6 address 2001:4860:800b::6a

        If you actually try to connect to 208.67.219.132 you end up at "hit-nxdomain.opendns.com" so it looks like there are some DNS shenanigans going on at your end.

        In any event, you can't get to the actual Google website with its "bouncy" logo unless you do so over IPv6.

      • by nweaver (113078) on Wednesday January 27, 2010 @06:44PM (#30926520) Homepage

        You got trapped by OpenDNS. OpenDNS is VERY agressive at wildcarding network failures:

        132.219.67.208.in-addr.arpa. 18794 IN PTR hit-nxdomain.opendns.com.

        So even though there is a valid name for ipv6.google.com (the Google DNS servers return a valid reply with a 0-size answer for an A query, and the whole data for an AAA query), OpenDNS instead goes "hey, lets wildcard it and return our server!"

        This behavior is why I'm NOT a fan of OpenDNS.

    • by Matt_R (23461) on Wednesday January 27, 2010 @07:24PM (#30927194) Homepage

      # host ipv6.google.com
      ipv6.google.com is an alias for ipv6.l.google.com.
      ipv6.l.google.com has IPv6 address 2001:4860:c004::68

      # host www.google.com
      www.google.com is an alias for www.l.google.com.
      www.l.google.com has address 66.102.11.99
      www.l.google.com has address 66.102.11.104
      www.l.google.com has IPv6 address 2001:4860:c004::68 :)

  • by Orbijx (1208864) * <slashdot DOT org AT pixelechoes DOT net> on Wednesday January 27, 2010 @06:14PM (#30925962) Homepage Journal

    If Comcast actually does what they're saying on the tin, maybe the other ISPs will follow suit.

    This just might be a good thing.

    • by jonbryce (703250) on Wednesday January 27, 2010 @06:23PM (#30926134) Homepage

      Maybe, but it is a difficult sell to customers. They will want to know what ipv6 enables them to do that they can't do at the moment. Being able to visit ipv6.google.com and do exactly the same things that they can do on www.google.com at the moment, and being able to see a dancing turtle at www.kame.net isn't really going to seal the deal.

      • by david_thornley (598059) on Wednesday January 27, 2010 @06:35PM (#30926360)

        Except that, if this relies on customers making a decision, it's dead.

        Modern computers support IPv6. Modern consumer-level routers don't necessarily (mine doesn't), so the connectivity provider needs to provide and/or recommend equipment that does. Provide connection instructions that start up both IPv4 and IPv6. Leave the customer out of it, since 99% of customers don't know what IP is in the first place.

        • by japhering (564929) on Wednesday January 27, 2010 @06:44PM (#30926534)

          Except that, if this relies on customers making a decision, it's dead.

          Modern computers support IPv6. Modern consumer-level routers don't necessarily (mine doesn't), so the connectivity provider needs to provide and/or recommend equipment that does. Provide connection instructions that start up both IPv4 and IPv6. Leave the customer out of it, since 99% of customers don't know what IP is in the first place.

          If the customer really, really wants to know what is the advantage for him.. the simple answer is continued access to the internet.

          • by petermgreen (876956) <plugwashNO@SPAMp10link.net> on Wednesday January 27, 2010 @08:02PM (#30927630) Homepage

            IMO the most likely endgame for IPv4 is the removal of public IPs from home lusers (replacing them with ISP level NAT) to give them to more profitable buisness customers.

            Continued ability to use peer-peer stuff efficiantly will be the main selling point of IPv6 to home users, of course peer to peer is something ISPs want to strongly discourage.

            I really really doubt that most ISPs would dare cut a customer without v6 supporting network equipment. computers and software off from the internet at this point. Not to mention that most websites aren't available on v6.

            One complication is that iirc some of the largest American cable providers have actually already run out of private v4 IPs for thier internal network and so are putting cable boxes on public IPs. If they could move the cable boxes to v6 they could reuse their v4 addresses (whether public or private) for peoples internet connections and thus avoid the need to partition thier network into mulitple addressing domains.

      • by sjames (1099) on Wednesday January 27, 2010 @06:48PM (#30926604) Homepage

        Are you on the same internet I am? The internet that went crazy for the "I kiss you" guy? The same internet filled with people who will sit through a sales pitch to get a $0.10 blinking LED toy?

        IPv6 MUST be the best thing ever! It's like the internet PLUS TWO!

      • by nnet (20306) on Wednesday January 27, 2010 @06:49PM (#30926628) Homepage Journal
        There is no "selling point". The move to IPv6 will be transparent to Joe Sixpack pr0n downloader/web browser/emailer, and Grandma Moses. The move is required in order for them to stay in business, and provide services to their customers. Its that simple.
      • by owlstead (636356) on Wednesday January 27, 2010 @07:01PM (#30926862)

        It allows multiple clients to have their own IP addresses. Which means that you don't have any limitations you have with IPv4 while hosting stuff (bittorrent, games). If your router supports IPv6 of course, but I don't think that network appliances are the problem. Things like mobile devices (for which IPv6 would be great) are more likely to suffer because of lacking IPv6 support.

        • by tftp (111690) on Wednesday January 27, 2010 @09:30PM (#30928450) Homepage

          Which means that you don't have any limitations you have with IPv4 while hosting stuff

          It also means that ISPs can now charge you per computer, instead of per IP (that you then NAT to cover your whole house.)

          If your router supports IPv6 of course

          I have three routers, none of them support IPv6, and without specifically searching I don't know any that do (except Airport.) Often it's hard to tell even holding the box in hands at the store.

          I don't think that network appliances are the problem.

          Unless, of course, you have such appliances. There are millions of devices [lantronix.com] that are IPv4-only. Support for IPv6 just started, and there is zero chance that earlier products will be upgraded (they are out of warranty by now.)

          Things like mobile devices (for which IPv6 would be great)

          Mobile devices (cell phones) are self-contained, so they are welcome to have whatever IPv$x they want. These devices are not a problem, and they indeed benefit from IPv6. However everyone else, industrial and residential PC and gadget users, will be in need of some serious 6 to 4 bridging. There are just too many embedded devices which are IPv4 only *and* out of maintenance. We also should remember that majority of network-aware applications are IPv4 only. This is even more true in niche, expensive applications, those that use networked license servers, for example. In their market even if a newer app is available and supports IPv6 you have to buy it again; support on that level is not free.

          • by TheRaven64 (641858) on Thursday January 28, 2010 @08:26AM (#30931862) Journal

            Having IPv6 doesn't mean that your v4 devices are going to stop working. They'll still be able to make outgoing v4 connections even when every consumer network is double-NAT'd. Most of these can then continue functioning via gateways even if the rest of the world is v6-only. You can, for example, run an IPv4 web proxy which forwards connections to IPv6 web servers. I don't think there are consumer-grade things that do this yet, but you can already get routers that handle external NAT, so IPv6 addresses returned by DNS are mapped to a private range (usually 10/8), then connections to these addresses are forwarded to the real server. You can also get reverse NAT devices that sit in front of things like v4 networked printers and let people talk v6 to the gateway, which then talks v4 to the printer. These are trivial to implement because everything above the IP layer (TCP or UDP, then IPP, HTTP, or whatever) is the same.

            You act like this is some kind of new experience for the Internet, but back in the '90s, there were AppleTalk, IPX, DECnet, and various other protocols deployed on local networks, all connected to the Internet via gateways. You'd run the same high-level protocols (SMTP, HTTP, and so on) on the local network as on the Internet, but different layer 1-3 protocols. Your mail client wouldn't know or care that your OS was talking AppleTalk to the gateway, the gateway was talking IP to the remote gateway and the remote gateway was talking DECnet to the mail server.

    • by N7DR (536428) on Wednesday January 27, 2010 @06:55PM (#30926744) Homepage

      I was part of the team that wrote the IPv6 portion of the DOCSIS 3.0 specs. Although DOCSIS 3.0 added a huge number of features, the two that the cable companies were most desperate for were channel bonding (so they could compete with fiber) and IPv6 support.

      IPv6 has been internal testing with major cable operators for several years now. Comcast was always likely to be the first to deploy it (for reasons that I can't go into) but I expect the other major operators to follow suit within a year or two.

  • I'm impressed that Comcast is talking about it trials publicly and engaging customers. Many service providers run stuff in private, don't tell their guinea pigs, I mean customers that they experiment on, and then just select whatever seemed convent for the service provider. Engaging people in a trials like this, seems win/win for the customers and service providers.

  • most routers? (Score:3, Insightful)

    by arbiter1 (1204146) on Wednesday January 27, 2010 @07:06PM (#30926926)
    I guess it means most companies (aka dlink, linksys, etc) have to get off their ass and add support to their routers since most home routers don't support ipv6.
  • by rritterson (588983) on Wednesday January 27, 2010 @08:09PM (#30927698)

    For what it's worth, I signed up for the trial. Despite the level-1 tech support's crappiness, and the relative overpricing of their services, Comcast's network department does a pretty good on the backend. Our area has gone from 3mbps to 16mbps (with a 50mbps tier available) in 8 years, and has already completed the analog reclamation process in our area. Good on them for getting a head start on IPv6.

    I presume they are going to want to do end-to-end IPv6 eventually, instead of assigning a single IPv6 address to my modem, and then continuing to use IPv4 NAT behind it. However, if they are going to do that, several things are going to have to change:

    1. Router default settings will have to change. Out of the box, most home routers use NAT by default, and, since most people don't change the settings (based on the number of 2WIRE### SSID's broadcast to my house), they'll have to redo them for IPv6.
    2. Auto discovery services will have to get better. I can say, categorically, that OS X is better than Windows and Linux at automatically finding nearby machines and devices that do not have a static IP/DNS A record assigned to them. The other 2 OSes will have to catch up, because, while a quartet of triplets is annoying but manageable to type, an IPv6 address will be a bear to copy down.
    3. A debate between static and dynamic IP addresses will have to take place. Ideally, a device would get a static IPv6 address assigned to it and keep it forever, no matter where it roamed and went. It'd be akin to a routable MAC address. However, if we do that, we'll run out of IPv6 addresses more quickly (though still not fast), since things like phones get recycled fairly frequently. But there are several obvious downsides to continuing to use totally dynamic IPs.

    Finally, as an aside, it's interesting to me, at least, how Apple Airport Base Stations do IPv6 routing automatically via a tunnel provider (as another commenter noted). Comcast doesn't support any IPv6, but when I'm connected to my router at home I get full IPv6 support transparently. Apple doesn't even mention this as a feature on the box, and it's not highly configurable either. So why did they spend all the effort to get it that way? Are they trying to stay so far ahead of the IPv6 curve no one will ever complain they're behind?

    • by JesseMcDonald (536341) on Wednesday January 27, 2010 @09:16PM (#30928356) Homepage

      I can say, categorically, that OS X is better than Windows and Linux at automatically finding nearby machines and devices that do not have a static IP/DNS A record assigned to them.

      That would be strange, since Linux uses exactly the same system as OS X (mDNS) for advertising local machines and services. You didn't disable the Avahi daemon, did you? It's generally enabled by default in new installations. You should be able to refer to any Linux machine on your local network as hostname.local, just as with OS X.

      Windows is a bit behind on native support, of course, but you can install Apple's Bonjour for Windows [apple.com] software to get the same effect.

    • by TheRaven64 (641858) on Thursday January 28, 2010 @08:51AM (#30931992) Journal

      Ideally, a device would get a static IPv6 address assigned to it and keep it forever, no matter where it roamed and went

      Why? What problem does this solve? You should be advertising machines via DNS, not by their IP address. If you move to another network, you update the DNS entry. If you're talking about mobile devices roaming between networks then I suggest that you look at Mobile IPv6. This uses IPsec (optional in IPv4, a required bit of IPv6) to update the routing tables when the machine migrates. If you have a Mobile IPv6 address, you can move the machine between networks without dropping connections. Making this the default would be silly though; how often does your PVR move between networks? Your web server? Even my laptop doesn't usually need to maintain connections when it hops between unrelated networks; I suspend it and resume it in between, so the connections would drop anyway. Of all the devices that I might own, only a pocket computer / telephone would actually need this.

Behind every great computer sits a skinny little geek.

Working...