Some industry analysts are saying that edge computing has the potential to be a larger and faster-growing market than public cloud currently is today. (I have no vested interest one way or the other but that seems like a really big market.) They claim that many of the applications and services we use today in public clouds are not suited to the long round trip times to get from our devices, across one or more service providers, to a public cloud and back. The business use cases of edge computing are farming, manufacturing facilities, and automotive real-time systems. Regardless of the business use cases around edge computing, one thing from a technology standpoint is demonstrably clear: Edge computing needs IPv6.
One of the key characteristics of edge computing is that the compute function will likely be a shared, distributed resource. It will be co-located in mobile cellular towers, in small shopping centers or retail shops, or even along highway stops. An example of this is Central Office Re-architected as a Data Center (CORD), which leverages legacy central offices geographically close to the residential subscribers for modern data center architectures. But being able to provide this distributed computing resource to many providers and customers will require that each compute unit be uniquely addressable. Thus, IPv6 is a natural fit. Each unit of compute can have a unique (and even the one-time use of an) IPv6 address to allow a customer, SaaS application, or cloud provider to manage that compute unit. In fact, many providers will likely choose to run IPv6 overlays to give that compute unit an IPv6 address from their own IPv6 allocation, making security and compliance easier to achieve and manage.
Not only is the abundance of IPv6 addresses a strength in this situation, the fact that mobile networks today are often IPv6-enabled (and in some cases, like T-Mobile in the US and Reliance Jio in India, IPv6-only) means that IPv6 will play a critical role in making the resources in an edge solution available on the network. So, the logical next question is, if IPv6 is critical at the edge and available in many public cloud solutions, what do I need to do to be ready? Adopting IPv6 so that your organization can use these edge solutions and tie them back to your public cloud provider would be at the top of my list of recommendations. For larger organizations, it will likely take a couple of years to adopt IPv6 in a substantial way. So, if you suspect your business is going to have edge computing as part of its strategy, then IPv6 must be on your short-list of necessary-to-adopt technologies.
Will it still be possible to do everything with IPv4? Of course, the edge providers are going to figure out how to do that for you (likely for a higher price!). They will end up making use of or SIIT or SLB46 to accommodate your lack of IPv6 adoption. But your organization will be at a distinct disadvantage compared to those enterprise networks that have native IPv6 access. The state and brittleness of the translation services, plus the reliance on your edge partner to keep those services running, introduce risks that over the long term can and should be avoided. By adopting native IPv6, even in a limited capacity, these risks can be minimized or avoided altogether.
So, what technical advantages do companies get by embracing IPv6 for edge? There are some characteristics of IPv6 that support the edge compute model. For instance, traffic engineering, service chaining, software-defined networking, and network function virtualization are all existing IPv4 technologies that carry over into IPv6, so it can leverage those. But from a unique value trait, Segment Routing IPv6 (SRv6) and extension headers are the standouts. These additional characteristics give IPv6 the ability to provide more options and capabilities than IPv4 and provide areas of innovation for edge.
More importantly, the business flexibility to create new and interesting solutions given the flexibility of having IPv6 on those compute hosts will enable new innovation. Moving back to allowing peer-to-peer in an edge environment means it will be possible to avoid public cloud bottlenecks for many services and transactions. It means your solution can run autonomously, without the need for data, control, or management traffic having to be centralized through a data center. Also, allowing SaaS and public cloud providers to become the gateway and/or proxy for government surveillance and control of data of most user Internet traffic is a real privacy concern. Edge computing may help mitigate that issue.
So how are you planning to adopt IPv6 so you can leverage public cloud, edge computing, and give your distributed workforce access to the entire Internet? I encourage you to get started with your IPv6 project and let us know how it is going!
You can find me on twitter as @ehorley and remember…
IPv6 is the future and the future is now!