APNIC has warned that the internet will run out of IPv4 addresses in late 2011. While there's still time to deploy IPv6, a group of telecom insiders said that the sector pushing hardest to get v6 out the door - the mobile guys - is also the furthest behind in IPv6 readiness. Participants in an online discussion, moderated by Telecom Asia global technology editor John Tanner and sponsored by Cisco, also raised concerns about service providers' OSS/BSS platforms, which just like in any network project can always catch you out.
Telecom Asia: We've known about IPv4's limited address pool for a long time, and IPv6 has been around since 1999. Why has IPv6 take-up been so slow?
Hon Kit Lam: I think we all understand that NAT (network address translation) is one of the factors delaying deployment. The other factor is device readiness for v6 - support for v6 from the OS and the device. But that will not be a factor for much longer, because we are already getting more OSs and devices supporting v6, and with more mobile devices needing an IP address, that will drive more demand for v6 support.
David Woodgate: There's been very much a chicken-and-egg situation about vendors waiting for sufficient demand for IPv6 and customers waiting for vendors to support IPv6. We're getting to a point now where the vendors are supplying equipment that is IPv6-capable across the board. Plus it has taken a long time for people to be convinced that the addresses were actually going to run out. I think it's also worth remembering that we're talking about roughly 15 to 20 years here since the development of IPv6. That sort of compares with how long it took for IPv4 to become prevalent.
Chris Liljenstolpe: I would add that a lot of carriers have a finite set of resources to invest in their networks, and a large portion of those resources are driven by one of two things: a crisis in the network, or customer demand and customer feature requirements. And if customers aren't asking for v6, it becomes hard to push your investment portfolio toward deploying v6.
Fred Baker: It's going to take time and resources. If we have to do it all at once, it's a lot harder. At the IPv6 Forum we've been trying to suggest to people that they start trial deployments early, that they work on their OSS software early. So, it's really not a matter of us saying every year, "You need to deploy IPv6 and this time we really mean it." The statement has been, "It's going to take time to test things and get it right, so start."
Tanapon Chandavasu: What we hear from our corporate customers who have tried our v6 service is that you don't see a lot of websites that are v6 ready. So they don't see it as an urgent requirement. I was happy to hear that Google and YouTube now support v6. But I don't hear anybody else coming out and saying that. So I'm still worried.
Tsao Chih-cheng: One problem we see is cost overhead. Upgrading the network infrastructure and equipment for IPv6 applications will increase investments and costs. That is a big negative factor for return-on-investment consideration.
Jeff White: I've had a number of conversations with CXOs where they've characterized it as the "IPv6 tax", if you will. Because the other side of this is, if you look at the economics of it, how do you end up monetizing the investment that you have to make around IPv6? Am I really able to drive incremental revenue through IPv6 deployment? Mobile operators have been talking up the need for more IP addresses for mobile devices for some time. How ready are mobile networks to support IPv6?
CL: We're advancing our v6 plans across multiple networks, one of those wireless. However, if I look at the end-node IPv6-ready state, I would say it's the furthest behind in the wireless world. In the mobile space, there's very few platforms, and none of the ones that are "painful" from an address-utilization standpoint are v6-ready. And the interesting bit is, it was the mobile guys who were pushing really hard to get v6 out the door, but now it's those very same end-nodes that are not ready to consume an IPv6 address or use an IPv6 interface.
Mallik Arjun Rao: From a mobile operator perspective in India, the IPv6 requirement will primarily be driven through mobile once the 3G launches happen next year. Most of our core, including MPLS, is IPv6-ready, because all these nodes have been built in the last 34 to 36 months. So most of my infrastructure is capable of IPv6.
Fred, does the mobile industry have enough time to get everything sorted out?
FB: I think that they probably have time, but they need to get their behinds in gear. From people that I know, it sounds like Nokia's a little bit further ahead than some of the other 3G vendors, and I know that they've all got a lot of plans in place to do stuff. But you're right to say that they're behind.
What are the top challenges that still need to be addressed in order to get IPv6-ready?
FB: We just did a survey in the IETF of ISPs and other people who are running such networks. The comments we got back was that those who believed that there was an issue with v4 were moving to v6, and that there were ISPs that believed that we would never run out of IPv4 address and had absolutely no plan for it. And that remains true right now. Those who want to bet that way, I think they're silly. But it's their bet. Somebody has to go out and do some education and convince people.
TC: One issue for us is that a lot of people within my company think that the NAT gateway will be able to prolong and extend v4 so that we don't have to really deploy v6 on every front right away. The problem right now is that NAT doesn't work 100%. We know for a fact, when we deploy, 10% of our customers are using applications that NAT cannot support or will not support. We get complaints from the call center asking us to fix some of these problems already and we can't. So I've been telling them that NAT is not the way to go, but there are still people within my company that think deploying NAT will work.
MAR: The issue for us is the timing: When do we really start seeing and thinking about it then getting it deployed? When do we really see the implementation happening?
CL: Internally, my biggest concern is probably the OSS/BSS systems, just like it is for any network project that has nothing to do with whether it's v4, v6, a new platform - it's always the OSS/BSS that catches you out. The other big one is the lack of a good solution for the legacy v4 nodes on the client side of the network. I'm not so much worried about v6- and v4-only content. My concern is the v4 nodes in someone's home: the media player, etc - the consumer electronics devices that are not going to get replaced for years. The fact that the industry is just now starting to address this is a problem.