News Stay informed about the latest enterprise technology news and product updates.

How to convert IPv4 to IPv6: A network architect’s view

Deciding when and how to convert IPv4 to IPv6, service providers face numerous technical and business issues; find out what matters most in this Q&A.

It’s easy to assume that every service provider has already converted IPv4 to IPv6 networks, especially because IPv4 addresses really are running out. Best not to assume, however. Moving a network from its native IPv4 to IPv6 isn’t simple, even for the most skilled network engineer, and business implications come into play along with the technology issues.

What’s so complicated about the IPv4 to IPv6 transition? For starters, most service providers deploy equipment from multiple vendors in their networks, which makes it time-consuming to convert IPv4 to IPv6 in terms of upgrading code. After the hardware and software issues have been addressed, providers have to build an IPv6 overlay across the network without breaking it. Then there’s testing and troubleshooting. Finally, let’s not forget that providers with IPv6 networks will still need to serve customers with only IPv4 connectivity for years to come.

To find out how service providers are doing with the technology and business issues that affect the IPv4 to IPv6 transition, talked to Brandon Ross, network architect with the Torrey Point Group, an experienced network engineer who has worked with IPv6 for years. Ross recently spoke to service providers at the 2011 IPv6 Summit about the pros and cons of converting various IPv4 address allocations into IPv6. In your view, are service providers ahead of the curve or have they waited too long to convert IPv4 to IPv6?

Ross: I’m very much a capitalist sort of guy, so my answer is that service providers will tend to migrate to v6 at just the right time on average. This means there’s a lot of frustration among technical people that providers haven’t migrated sooner. But that’s not really how business works. It waits until there’s some sort of turning point in the economics before it makes sense to deploy a new service, and v6 is a new service like any other.

Certain service providers are ahead of the curve— and will definitely benefit.  Hurricane Electric [an ISP with a specific focus on IPv6] is a great example of that. It really made a name for itself in IPv6, and it will probably be rewarded for it.

Providers that end up waiting to be one of the last to migrate may actually be saving a lot of money on their infrastructure costs because they’re not leading-edge; they don’t need the latest equipment, and they can keep reusing those old routers and old customer aggregation devices and CBE for years to come. So there’s a mixture. I think only the market can really tell whether it’s too early or too late for any particular provider. Do you work with many service provider clients on IPv6 issues in your work with the Torrey Point Group?

Ross: Absolutely. Our most common engagement around v6 has been basic v6 deployment on top of an existing v4 network. A lot of our clients have a v6 deployment that they need to do, but often they don’t have the staff or the time to develop the architecture from scratch. We help them by creating it and working them through the process so their staff gets experience.

Other problems are more specific. We help people who need to figure out which version of software they need to use to upgrade their routers to support the features they need in v6. We may take that and do some testing around it, or help them with their addressing plan or acquire v6 addresses from one of the registries. What’s involved in converting IPv4 to IPv6 from a high-level perspective?

Ross: One thing to figure out is how to deploy IPv6 addressing on your network without redoing your addressing plan. It depends a lot on what kind of services you need to offer on your network, how big your network is, how it’s laid out, and most importantly, it depends on how your existing IPv4 addressing plan is laid out.

But the benefit is that if you can do it, then you can really reduce the amount of documentation you need to do [it] because you automatically know that a particular IPv4 address transforms into a particular IPv6 address without having to look it up in a table. Where are most service providers in terms of converting IPv4 to IPv6 addresses?

Ross: I think most are in the middle ground. The middle ground is that most have a deployment in IPv6, but they are not in a position where they can automatically offer it to every single customer, although they usually can when asked. Usually much, if not all, of the larger ISPs’ infrastructures—at least the larger ones—support IPv6, although there are certainly exceptions to that. How many options do service providers have in terms of making the transition from IPv4 to IPv6 addresses?

Ross: From a high level point of view, not a lot. With almost every provider, it’s going to make sense to deploy just a standard dual-stack environment. Then again there’s not a lot out there that they would be running similarly to MPLS.

You could provide v6 services to your customers without necessarily supporting v6 on every one of your routers by creating a tunnel back to some routers in your network that support v6. I suspect that a good number of providers will end up doing it that way to avoid having to roll out v6 across their entire network. I think the majority are or will be dual-stacked in a pretty short order. What is the best way for service providers to convert IPv4 to IPv6?

Ross: That depends a lot on the provider. From a pure engineering perspective—dual stack is definitely the cleanest and the 'right way to do it.' You have to look at the economics, though. For a lot of providers that may have older equipment, that would cost them significant capital to replace because it doesn’t support v6. It would make a lot of sense to do an overlay or some sort of compatibility approach instead of deploying v6 across their whole network. What are the biggest pitfalls when converting from IPv4 to IPv6?

Ross: The biggest problem is moving too fast when you do convert IPv4 to IPv6. That really plays out in deciding whether you’re too early or too late to the game. The biggest problems service providers find is not usually what technology they chose, but doing it poorly without thinking it out in advance and not doing adequate testing. That’s generally not true with the larger providers. It’s generally a problem with the smaller ones who have fewer resources to really flush out what they’re doing.

There are always bugs; there are always going to be problems, and people will always run into issues. And there are always going to be choices in how you approach the problem; but really, your methodology of rolling out new technology, regardless of whether it’s to convert to IPv6 or not, is really the key factor. Is the pressure on the service provider to make sure enterprises move to IPv6?

Ross: I don’t think it really is on the service provider. I think it comes back to economic factors, where your enterprise customer comes back to you, as a service provider, and says,  ‘I need 200 more IP addresses.’  At some point the service provider is going to have to say,  ‘Well I’m running really short on them. I’m going to have to charge you a lot more than I used to.’ It’s those kinds of factors that will naturally push enterprises to eventually look for other options like using IPv6. Unfortunately, it also pushes them to other options like network address translation (NAT), which is generally considered less desirable. But I think that’s how the transition goes; I don’t think it’s a marketing issue as much as it’s a financial one.

Let us know what you think about the story; email: Allison Ehrhart at

Dig Deeper on Telecommunication networking