The ideas behind IPv6-only
I wrote in my 2017 predictions post that we would see early IPv6-only data center solutions this year. I wanted to dive a bit deeper into why this prediction is on my list and some of the rationale behind why this is starting to happen.
The current state of affairs for IPv6 adoption is really focused on initial adoption and on dual-stack solutions. So much so that we have, in many ways, lost sight of the end goal that we are trying to achieve by introducing IPv6 in the first place. The end goal is to move the entire public Internet to IPv6 to allow for expansion and growth while utilizing a protocol that has no foreseeable natural limitations on constraining that growth at all.
It seems we humans are creatures of habit and slow to adopt or move without significant motivation. As a result, IPv4 has stuck around far longer than its useful lifetime. Realistically, we have engineered solutions to extend lifespan of IPv4 and we are now paying out in considerable amounts the technical debt that those solutions entailed.
So how do we move forward?
Several milestones have been achieved in the move to IPv6 that could impact many of the decisions being made by large content networks and mobile providers that drive the argument that IPv6-only might make sense. Let’s review:
- IPv6 is preferred and used by every major mobile provider in the US
- IPv6 content is now available for 19% of the top Alexa sites
- IPv6 performs better for many of the major content providers such as Facebook, Google, YouTube, Netflix and others
- IPv6 is one third of the reported overall Internet traffic protocol for the U.S. today
- Android and iOS both have robust IPv6 support and transition technology capabilities
At some point, when evaluating the cost of building, deploying and operating a large scale data center for content that is consumed by mobile subscribers it become apparent that IPv6 is incredibly important. Then it becomes a cost exercise in determining are you better off operationally running a single protocol network or a dual-stack one. There are many reasons to choose a dual-stack network but cost and efficiency is not on that list. At a certain point, the burden of running a dual-stack data center becomes too high and we end up with either an IPv4-only or an IPv6-only network. Given that the mobile market adoption of IPv6 is significantly higher than any other (Verizon for instance is over 70% native IPv6 traffic on their mobile network) it becomes clear that moving to IPv6-only is the right path.
Some organizations have already started down this journey. They have arrived at this decision because it is best for their business or it is the most logical option given the parameters of what they do as a business. For example, power providers leveraging IPv6-only mobile solutions due to the number of end node power devices they have to manage. Regardless, the outcome is they are invested in IPv6-only networks and data center environments.
Let’s examine a few who started early. Facebook is operating IPv6-only data center and providing IPv4 as a service as an overlay to address those times when some sort of IPv4 is still needed. Paul Saab and others at Facebook have presented on this at several International and National IPv6 conferences. In addition to Facebook is the work by Tore Anderson who works for Redpill Linpro based out of the Netherlands. They consult to service providers focused on content and hosting. They chose to leverage a different transition technology to allow IPv4-only hosts to access and use the content from their network. They implemented Stateless IP/ICMP Translation Algorithm (SIIT) (RFC 6145) to make IPv6-only hosts reachable from an IPv4-only host. While the solution addresses the shorter term IPv4 needs it is hopeful they won’t be required to keep their services running.
As we rapidly approach greater than 50% utilization of IPv6 on the public Internet it will be up to the major enterprise networks to decide how to move forward. If they remain on IPv4 and only add IPv6 they will pay the burden and overhead of running a dual-stack network. Depending on that organization they may not be able to staff and operate a dual-stack network without significant investments. If they choose to move to IPv6-only then they have the structural advantage of a single protocol but in addition they are skating to where to puck is going, not to where it was (i.e. Wayne Gretzky). This is important for technology decision makers. Do you just continue with the status quo and potentially get leapfrogged or do you pivot and adopt the protocols and technology of the Internet of the future?
There is no perfect answer but I do feel that many companies will start looking at the situation dispassionately and start to realize that a move to IPv6-only can be a strategic advantage. It helps larger companies solve key problems like mergers and acquisitions, hyper scale growth and expansion, cloud scale, mobile software markets, direct customer engagement and security all by simply choosing to use IPv6-only.
So I am going to borrow the LinkedIn IPv6 team shirt motto and ask everyone to join the IPv4 disposal team!
You can find me on twitter as @ehorley and remember…
IPv6 is the future and the future is now!