9 Comments

Summary:

It’s almost World IPv6 Day, a time when ISPs and websites flip the switch to a different form of addressing system and very little actually changes for anyone online. But here are 5 things that could go wrong if ISPs don’t make the move to IPv6.

stock-crowdglobe-e1297787156925

Wednesday is World IPv6 Launch Day, a time when ISPs, major websites and network engineers will permanently flip the switch over a different form of addressing system. For the most part, the IPv6 transition will go unnoticed and few will care about what is an arcane and important element of Internet architecture. But here’s where things may go wrong as IPv4 lingers.

At its heart the IPv6 addressing issue is exactly that… a big change to a new type of address. Every device that is hooked up to a network, be it your iPad or a Facebook server, has an IP address so routers know where to send the packets that make up your Facebook profile or your Netflix stream. Sadly, back in 1980s when the lords of the net were thinking up an address scheme, they used IPv4, which only allows for a 32-bit address and about 4.2 billion total IP addresses.

But darned if the growth of the web (all those servers!) and the growth of consumer devices (all those smartphones!) hasn’t caused the 32-bit addresses to run out. So since about 2007 people started pushing ISPs and websites to switch over to IPv6, which allows for 128-bit numbering system (or a total of 340,282,366,920,938,463,463,374,607,431,768,211,456 possible addresses). But because of a variety of workarounds put in place by ISPs and the general lack of hellfire and damnation to ensure ISPs and web companies make the change or risk getting cut off from the Internet, the transition to IPv6 is happening slowly.

Chart courtesy of the FCC.

And for the most part that isn’t a problem. But, as we consume more apps that take advantage of several ports (think of a port as a door that allows certain types of traffic through), and more people in our homes use those apps, consumers may feel a bit inconvenienced. This is a result of the workarounds that ISPs have in place to forestall their transition to IPv6 and ensure that customers can still access sites that haven’t already transitioned. But here’s how the delay in shifting to IPv6 and the reliance on a workaround could affect you, as a consumer according to a report from the Broadband Internet Technical Advisory Group. Bitag is a group of communications engineers and researchers created to help promote good Internet policy.

Trouble with iTunes or Google Maps: Certain apps such as Google Maps or iTunes use more than one port to communicate back to the service and because users would be sharing one or just a few IP addresses, those ports may not be available. From the report issued in March by Bitag:

For example, some applications, such as Apple iTunes and Google Maps make use of multiple ports for a single transaction. A single user running a single instance of an application on one personal computer may encounter no issues, while a family of four using multiple Internet connected devices concurrently could experience a situation where an open port is unavailable. Applications with no ports available for communication would not function as expected.

Security: When users share one IP address, the ISP generally creates an abstraction layer to determine where the packets need to go to in the home. But this abstraction layer becomes a security risk. By attacking one IP address, a hacker could take down or infect all Internet-connected devices in a home. It becomes a single point of failure.

Court orders and DMCA takedowns: Shared IP addresses can make it hard for an ISP to determine who is actually downloading copyright materials. This actually may not upset end users or the ISPs as long as no one gets dragged into court as part of a hunt for settlement dollars.

Pixellated YouTube videos: The workarounds associated with either sharing a single IP address among a block of users or even a block of homes is just one option. Another is running both networks simultaneously and translating traffic between them. This adds some computational overhead and latency that in CableLab tests caused there to be delays in streaming and receiving HTTP video packets.

Stories about IPv6: As long as networks linger in the land of IPv4 every June, it’s possible you will have to see headlines talking about the need to get everyone to transition to IPv6. Hopefully, ISPs and major web sites are getting on the ball.

Tomorrow’s IPv6 Day involves more than a dozen companies who have committed to permanently enable IPv6 (last year many of the providers turned it on but then turned it back off in an effort to understand what would happen when they switched). I’m personally encouraged that Time Warner Cable, my ISP, is finally announcing its plans to transition. Other major networks that have transitioned or are in the midst of doing so are Verizon, Comcast and KDDI.

  1. 6. State timeouts. Imagine you’re working on SSH and you’re idle for 5 or 10 minutes. Without fast keep alive packets, that session will drop. You actually see this a lot on mobile networks right now. Even if CGN/LSN implementations have the memory to keep long states, they can’t, as port exhaustion will be a huge problem.

    It should be interesting to see how this pans out for AT&T U-Verse when they deploy CGN this June (read dslreports story Karl did earlier this week).

    Geoff Huston did a great presentation last year at APNIC looking at why Teredo was so fail (eg. lots of breakage). The conclusion that he came across is NAT traversal and the fact there isn’t a RFC defined NAT implementation and every vendor has their own quirks that cause breakage. It’s a good read as to what failures might be like in an CGN environment. http://www.potaroo.net/ispcol/2011-04/teredo.html and http://www.potaroo.net/presentations/2011-02-23-dualstack.pdf

    Share
    1. Will, excellent additions to the article. But at this point even is there was a set NAT standard would ISPS deploy it, or has this horse left the barn?

      Share
      1. Stacey,

        The horse has left the barn for sure. If there were a NAT standard released today that dictated the exact behavior of a NAT, it would still take vendors at least a year or two to implement it. I would imagine that effort would be better spent deploying native IPv6.

        Share
    2. I’m surprised that a better reason of WHY IPV6 is now CRITICAL wasn’t given (for the benefit of lesser mortals than techies & geeks). Essentially we have now reached the point when NEW allocations of IPV4 addresses are impossible because the allocations have all run out! Now here is the critical piece of information which the ISPs have been ignoring for YEARS. ANY NEW IP address allocations that happen from here on out (pretty much anyway) will HAVE to be IPV6 or no address will be available at all. Anyone with such an IPV6 address ONLY for their web site, will be INACCESSIBLE by ANYONE at home on their IPV4 addressed PC as it simply will NOT be able to connect to that address. That is UNLESS their ISP has implemented a PROPER means of workaround. This means that HYPER URGENTLY, ALL ISPs will need to implement AT THE VERY VERY LEAST a means of providing FULL end to end IPV6 over IP4 (If you’re not technical, don’t worry – the ISPs are able to do this already BUT they HAVE to implement it – MOST HAVEN’T YET!!)

      Fortunately anyone using Microsoft Windows XP or later , or Linux, or an Apple MAC already has the necessary ability built in to handle IPV6 addresses WITHOUT NECESSARILY having to buy a new router – although LONG term that would be best, no argument). Without the ISPs moving VERY quickly however, it is highly likely that some new web sites coming onto the Internet over the next year or two WILL BE INACCESSIBLE to ANYONE on IPV4 whose ISP has NOT implemented the necessary steps to handle it.

      Bottom line is – KICK your ISP until it hurts (them) to get their commitment dates for IPV6 over IPV4 “tunnelling” (minimum) and full IPV6 implementation (which will include the latter) preferably and if they say “it’s not necessary yet” FIND ANOTHER ISP ASAP as the one you have is in denial mode and WILL cause you problems. The ISPs SHOULD have done this YEARS ago, but in fairness the major problem has actually been with the complete failure of the router manufacturers to get their heads out of the sand and fully implement IPV6 on their products MUCH sooner. I was working with guys who were discussing the move to IPV6 almost 20 years ago!

      Share
  2. So should I be hounding Verizon for more info / a new router, or do I just sit back and wait for magic to happen? I’m still very confused. I’ll probably call Verizon and sound like a total dink.

    Share
    1. Dave,

      It would be in your best interest to hound Verizon but I’m not sure it will get you anywhere.

      New router wise, I think it’s a pretty safe bet they will be doing DHCPv6-PD when they do finally deploy.

      Share
  3. IPv6, which allows for 128-bit numbering system is very nice!

    Share
    1. Hi Megabeet,
      Of-course the concept of IPv6 is really nice. It gives so many of combinations for providing a unique address to a node connected to network. Also, the count of combinations is so high that, the node addressing system can never be exhausted.

      Share
  4. IPv6 sounds like it will be a good thing. I don’t see why ISPs wouldn’t switch.

    Share

Comments have been disabled for this post