The road to IPv6-only in the Enterprise Data Center
Almost every Enterprise is operating some sort of data center today. While Cloud adoption is happening rapidly and the major providers (AWS, Azure and Google) are gaining workloads at an astonishing rate the clear reality is that it will take years (if not a decade or more) for this transition to manifest into some sort of equilibrium, likely resulting in an on-premises to cloud ratio of 55% to 45% of workload respectively. This means that a hybrid and more than likely multi-cloud world will be in your near future. The road to IPv6-only for enterprises will likely be driven by two major adoption pressures. The first is mobile (wireless) application workloads and the devices we are using. The second is mass service provider deployment or how people are connecting.
Let’s examine these to see how they impact your deployment of IPv6.
Mobile applications on next generation wireless provider networks have new standards and requirements. Every mobile network today in the United States provides IPv6 and several are trying to move to entirely IPv6-only. This move to IPv6-only is not random, there are some real-world business drivers pushing these providers down this path.
- They have run out of public IPv4 addresses or are close to depleting what they have left.
- The cost per public IPv4 address is sufficiently high enough now to have crossed or come close to crossing the threshold of too expensive. Current cost of an IPv4 address is around $11 to $15 per address on the open market. (This means it can cost you twice as much to obtain an IPv4 address as the compute unit, you can get a Raspberry Pi Zero for around $5 – crazy!).
- The cost to run large scale Carrier Grade NAT (CGN) solutions are getting too high.
- The cost to operate a network with dual protocols is growing in cost and complexity.
- Well-constructed mobile applications are indifferent to which protocol they operate with, nor should they with a few exceptions.
- Performance of IPv6 in many cases can be superior to IPv4 (though I am in the camp of thinking this will change when the greater than 50% adoption rate has been achieved).
- Many of the public services you will consume for both leisure and business will have IPv6 available.
Now that you understand the provider and mobile application side of the equation let’s balance out the other side and see what impact this might have on enterprise adoption.
What motivates an enterprise to adopt or select a given set of technology? Some adoption is around standards (E-mail, DNS, HTTP(S), SSH) which are clear examples of this along with the IPv4 protocol. It can be argued easily that the adoption of IPv4 by every company big to small has been one of the greatest standard adoption to ever occur (maybe the Metric System? Except those last 3 pesky holdouts). What is interesting is the fact that the other widely adopted standards have no direct dependency on IPv4, at least with current implementations. So why is the adoption rate for IPv6 so slow in enterprise relative to the wider Internet (remember the US adoption rate is north of 30% and growing)? The simple answer is that today, the motivations for an enterprise to adopt IPv6 and move to dual-stack are lacking. Their motivations for spending and investment are in other areas currently. Let’s put it this way. We have invested across the U.S. in narrow gauge railways and we have gotten widespread adoption. We are now telling everyone, everywhere that we must move to standard gauge railway. Worse, they are not backward compatible; they require that we touch every piece of track we own and doing this doesn’t provide any function difference for moving a train car from point A to point B.
You can see why so many enterprises are not interested in making the change. It isn’t like our collective governments are giving tax incentives for companies that do the work (something that might have to happen). What do we do now?
The answer may come from the fact that the standards of how others operate will force the change. While every enterprise might maintain a narrow-gauge rail line they eventually must interface with a port, transportation hub or another enterprise. As more of these moves to standard-gauge rail line it becomes more complicated to interface and keep things moving at the rate that is considered optimal without impact to the business and more importantly to the cost of doing that business. At this point, companies start doing the financial math and determining if it is more cost effective to adopt the new standard. From the behavior, I have observed as of late, it seems we are coming close to that threshold (at least here in the U.S.). Many companies are starting to find both advantages and specific use cases where IPv6 addresses business needs in a more compelling way than staying with IPv4. It may start with a small deployment to prove out the use case and then grow organically or it could be that someone did the full cost analysis and determined that adoption was far more cost effective over the lifetime of the service or business need. Regardless, the result is the same, higher adoption rates of IPv6 in these companies.
Mobile (wireless) and residential providers in the U.S. did the financial math years ago and determined IPv6 adoption was the only way to stay in business. They had to make the move to continue to have a product to sell. Many of the largest content providers did the same analysis and pushed IPv6 forward for their needs. The cloud providers have done the same and here we are, waiting for the enterprise and smaller businesses to adopt.
What is unique for the enterprise data center should do with refresh rate (faster than campus or WAN replacement) and the fact that many “data center” solution now incorporate a cloud (both public and private). If you are interfacing to a public cloud you now have the option of utilizing IPv6 and with your end users having IPv6 native on their mobile devices directly to a cloud service suddenly the need for IPv4 has gone away. This means you can adopt and use IPv6 for your on-premises private cloud and provide the same access, security, monitoring and controls just like the public cloud. It introduces the first case where considering IPv6-only could work end to end and help reduce complexity and simplify a deployment.
Not all companies are ready yet and many are not sure if they will ever get there because they feel their adoption of public cloud is far away. While it may feel that way now, the rates that are being published around public cloud adoption are astonishing and will likely push the wider adoption of IPv6 for those looking to build out new enterprise data center solutions. So, don’t be shocked if in your design and planning sessions for a new data center that IPv6 is one of the topics you need to cover.
Finally, if you are not familiar with IPv6 at all, the great news is that on April 25-26 at LinkedIn in Sunnyvale California is the North American IPv6 Summit which has a fantastic lineup of presenters ready to talk about IPv6. Many of the presentation are about practical IPv6 implementations, how they were done and the drivers behind them. The conference is very reasonably priced at $200 for a two-day event. So come join me and my fellow Infoblox IPv6 COE cohorts Scott Hogg and Tom Coffeen at the event, we would love to meet you in person, hear what you are doing around IPv6 and get feedback about our IPv6 blogs too!
You can find me on twitter as @ehorley and remember…
IPv6 is the future and the future is now!