A couple of years ago we wrote a blog post on service chains. At that time, NFV was still somewhat nascent, and the concept of stringing NFV services together was also in early stages. While some progress has been made in defining standards for service chaining (e.g., IETF SFC Working Group), there is still lots of work to be done before we have an industry-wide approach for how services are defined, orchestrated and managed.
Now, multiply the scope of NFV service chaining by a factor of, well, pick a number. Welcome to the world of edge computing and IoT.
The “new University model” of Edge Computing
You may be asking yourself “why is service chaining important at the Edge?” That’s a valid question if you think of the Edge in a traditional public cloud context. Most public cloud solutions utilize an “enroll/registration” model for edge computing and IoT. Think of it as going to college 20-30 years ago. You enrolled in college, then registered for your classes. From that point forward, your interaction with the college was pretty much “north/south” communications. Almost everything you needed was supplied by the college. Now, fast forward to today. You still enroll and register, but your interaction beyond that point is significantly different. There are a lot more “east/west” activities. You now depend on lots of external resources and peer groups. You pretty much must act in a somewhat autonomous manner, but at the same time, you need to be able to communicate with lots of other entities.
That is what the edge computing and IoT world are all about. Other than simple sensors which simply report “upstream” for data collection, the complexity of the Edge will be staggering. And simple enroll/register-north/south solutions will be inadequate for the vast majority of mission and life-critical Edge systems.
So how do we manage all this complexity? If we’re still waiting on standards for NFV service chains, how can we expect to deal with the Edge?
3 Essential Elements of Edge solutions
The answer lies in what we call closed-loop orchestration. I won’t go into the details here, but you can contact us and we’d be glad to share the complete picture with you. But for the sake of brevity, I’ll boil it down to three essential elements that are required for successful delivery of highly-distributed Edge solutions:
- Critical to the process is a manifest that describes what the solution is and what it needs to operate within a given set of performance parameters. The manifest should describe the “service”, not the solution. It should be independent of the underlying infrastructure. We use TOSCA (Topology and Orchestration Specification for Cloud Applications) as the preferred standard for describing applications and services.
- A topology model that discovers and describes the available infrastructure (physical and virtual) is key to mapping services to appropriate compute platforms. But topology by itself is inadequate. It must bring in the context of relationships, dependencies, and environment. Examples include latency, location, proximity, capacity, etc. Without these, services cannot be appropriately mapped to the correct infrastructure.
- Automated orchestration and provisioning are essential for the scope and scale of edge computing and IoT. We are operating in the age of the “machine internet” and traditional systems and humans are not capable of dealing with the complexity of service orchestration at the Edge.
There are other components that go into the creation of resilient Edge solutions. Things like security, policy, and software-defined networking. But again, for the sake of brevity, I’ll leave those to deeper-dive discussions.
Ultimately, delivering service chains at the Edge requires more than simply patching together a bunch of open source utilities and DevOps tools. It takes a platform approach that brings all the elements of service descriptions and topology and provisioning together in a seamless, automated fashion. That’s what we do over here at CPLANE.ai. You can reach out to us at CPLANE.ai if you’d like to learn how we can help build your next Edge solution.