Should IPv6 Have A Different Name?

A footnote to our earlier discussion on why IPv6 hasn't been widely adopted: the numbering of the newer version is clearly designed to suggest that it's an improved successor to IPv4. But is that name actually logical and helpful?

Dice picture from Shutterstock

During her keynote speech earlier earlier this week, Intel fellow and networking pioneer Radia Perlman (no big fan of IP as a standard generally) argued that the name was misleading. "IPv6 is not a new version of IPv4," she said. "It is a new protocol."

Perlman also pointed out that the version number was not actually utilised in a sensible way by IPv4. The version number is reported as part of a TCP/IP packet, but systems won't drop packets that are sent with a version number that doesn't match their implementation.

Perlman agreed with Geoff Huston's conclusion that we're not going to see a rapid shift to the newer format. "We've made IPv4 easier to live with, and the longer we live with it, the harder it gets to convert."

Lifehacker's World Of Servers sees me travelling to conferences around Australia and around the globe in search of fresh insights into how server and infrastructure deployment is changing in the cloud era. This week, I'm in Canberra for Linux.conf.au, paying particular attention to the systems administration mini-conference and sessions on virtualisation and best practice.


Comments

    I don't agree with Radia Perlman that IPv6 is a 'new protocol'. It is a new version of IP, with full upstack protocol compatibility. It only differs in its implementation, enough to make it incompatible with its predecessor IPv4.

    Arguing that the name 'IPv6' is misleading just because it's a significant implementation change from IPv4 makes about as much sense as arguing the name 'OSX' is misleading because it was a significant change from OS9. It's still the same product that accomplishes the same task, it just does a much better job of it, and in a different way to its predecessor.

    Call it the tubes?

    I haven't seen the keynote, so I may be wrong in my assumptions of meaning here.

    IPv6 can be seen as a 'new protocol' rather than a version bump due to lack of forward of backwards compatibility between the two standards. If a network device wants to support both protocols, it needs to run dual stacks or tunnel one protocol over the other.

    OS9 to OSX, many apps will still run without modification.
    If I want to use IPv6 at home, it's not going to work until for me until I replace my router.

      OS9 applications weren't compatible with OSX though. OSX was the first introduction of the Unix-based MacOS, and the only way to run native OS9 applications was to use an emulator, which is roughly analogous to tunnelling IPv6 through IPv4.

      At home you can use IPv6 as long as you run an OS with a tunnelling protocol installed. Windows since Vista has Teredo, and Linux has had 6to4 for a while. You don't typically need to do anything to run IPv6, your ISP just has to issue you an IPv6 address.

      Last edited 01/02/13 11:05 am

    Give it a cute wanky marketing name.
    What would you call it?
    Give some huge company naming rights or have a sponsorship deal.
    Like ZeroIP
    Nah. Too early in the morning to think of somethinh

      Okay, checking the trademark database, the only names available are... "Popplers" or "Zittslers"

    A dual-stack node IS backward compatible with IPv4-only. It includes full IPv4 capability. I can run any IPv4 application on such a node without any modification. An IPv6-only node is NOT backward compatible with IPv4-only. SomeIPv4-only network aware applications need minor modifications to work on it. An IPv6-only node will work with much legacy physical layer and link layer things (NICs, cables, layer 2 switches, etc). You will need to upgrade (or maybe just IPv6-enable and configure) other infrastructure devices (routers, firewalls, CPE, DNS, network managemnt, etc). IPv6 was based HEAVILY on the IPv4 standard, many IPv6 features are highly recognizable to an IPv4 engineer, but were implemented slightly differently (e.g. address resolution). Accessing IPv6-only servers from IPv4-only requires translation (ideally at layer 7 with a proxy - layer 3 translation still has many problems). I can make all incoming and outgoing web traffic in an IPv4-only (or IPv6-only) network dual stack overnight, requiring only a proxy server and minor changes to DNS. Ditto SMTP. Many apps require zero changes (e.g. most web apps in PHP, ASP, ASP.Net, etc - if the underlying web server and OS support IPv6). I do not need a new CPE if something inside (maybe just software on one node) has a tunnel endpoint. I've been running native dual stack inside my home (and work) network since 2005 and there is not a single ISP in my country that can provide even native (or even tunneled) IPv6 to me. I tunnel it from outside the country (in my case using 6in4). I happen to have a firewall that contains a tunnel endpoint, so internally there is no meaningful difference compared to having native service from an ISP (20 bytes extra overhead on each IPv6 packet, plus occassional additional packet fragmentation if something doesn't allow for this in MTU). It will be a while before there are many IPv6-only nodes, networks or servers, but the day is coming - IPv4 is End Of Life in many ways, not just availability of public addresses. NAT was a horrible kludge that was a necessary evil while IPv6 was being developed. IPv6-only networks are more robust, easier and cheaper to manage, etc. I see a day soon when many networks go IPv6-only and access legacy sites only via translating proxies. The first major example with be Internet via LTE advanced.

    To the point of the article, I think it is counterproductive to concentrate on IPv6 (or even come up with a cute new name for it). Most people never interact with it directly, and don't really care about (yes, I know - shocking). I prefer to talk about "The Second Internet" based on IPv6, that WILL offer radically new and better functionality (largely because of no NAT and working scalable multicast). THAT gets people's attention. People will be willing to pay additional money to get access to it. They will of course need some kind of IPv6 (native, tunneled, translated) to get access to this new Internet. Once this is in place, most will forget about IPv6. Many will never know it is even there. BUT - they will be able to do many cool new things because it is there. Few people understand the technical differences in SDTV and HDTV in terms of the signal being broadcast, or even the bandwidth required for each. But let them see a 55" screen with a tack sharp image, and they are willing to shell out bucks on the spot. Content providers have to be aware of it in order to provide content and services over IPv6 - just like HDTV. Perhaps everybody BUT the engineers who are making it happen should shut up about IPv6 and promote the Second Internet (see www.secondinternet.org). Most people will sit up and say "NOW you're talking!"

Join the discussion!

Trending Stories Right Now