What’s in an edge computing platform?

The term “platform” often refers to many different things and becomes even more confusing when linked to a new, not well-defined domain such as edge computing. We seek to clarify what an edge computing platform is and provide examples of the edge platforms being developed today.

What is an edge computing platform?

An edge computing platform is a horizontal piece of software that enables the automated deployment, update and management of distributed applications. It is similar to a cloud platform, though there are some key differences that we explore in the article below.

“Edge computing” and “platforms” are two of the most confusing and misused terms in the industry today; “edge computing platform” is exponential confusion.

For the purposes of this article, and wider debate, we will define edge computing as “processing capabilities closer to the end user/device/source of data, on physical compute infrastructure that is positioned on the spectrum between the device and the internet/hyperscale cloud” and platforms as “a software environment that is used to write and run software applications”.

Reminder: we already have cloud platforms

Before we discuss the features that make an edge computing platform, it is worth bearing in mind that there are benchmarks and standards already out there that edge computing is going to leverage and build on.

Although not all edge computing platform vendors come from the cloud world (e.g. some originated from industrial domains), many will use similar stacks as we have seen in cloud.

The Cloud Native Computing Foundation (CNCF) has played a key role in defining the cloud computing landscape, as well as providing a forum for reference code and test cases to ensure scalability for cloud applications. Although much of the focus has been on hyperscale cloud (private and public), their work has extended to hybrid cloud (of which edge is a component).

The CNCF landscape shows the different components that make up a cloud platform, as well as the companies in each category. The features outlined, such as App Definition and Development, Orchestration & Management and Provisioning, are all relevant in edge computing.

So, what’s new in edge?

The challenge is that the edge is not exactly like the (hyperscale) cloud; they differ in a number of ways. For example:

• Lots of disparate small edges: current cloud platforms are built for an environment where the underlying infrastructure is at least as large as a rack of servers – likely in the same physical location – whereas edge platforms will need to support infrastructure that contains smaller systems, is diverse and physically spread out.

• Distributed computing: it is unlikely that applications will run on a single edge, given that the edge is a finite (and potentially more expensive) resource. Therefore, workloads will be distributed dynamically across devices, edges and clouds.

• Traffic optimisation: to leverage an edge, traffic from an end-device needs to know which edge to go to in order to meet latency needs and ensure that edge is available and cost-effective to use. This requires an edge gateway and ‘edge cloud connect’ to optimise connectivity from device-to-edge, edge-to-edge and edge-to-cloud.

Interestingly, AWS’s Virtual Private Cloud does not support “edge-to-edge routing.”

This means that the edge stack is different to the cloud. Plus, there will be growing attention on the distributed cloud stack, which spans across different clouds: public, private, network edge, on-premises edge, device edge, etc.

The other nuance is that edge/distributed cloud applications are different to cloud applications. We discussed some of the needs of application developers for edge computing in our report What edge developers want from telcos now. Many of their requirements centre on being able to access edge clouds, having clarity on pricing models and being able to easily use platforms to ensure applications requirements are met (e.g. latency).

Examples of edge platforms today

We at STL Partners have started monitoring edge computing software companies, adding them to our Edge Computing Ecosystem Tool under “Cloud Infrastructure” and “Application/Software”. The diagram shows different edge computing platforms and “which edge” they are primarily focused on. Given that this is a rapidly evolving domain, if any of the information below needs to be updated, do let us know.

GSMA’s new 5G and edge computing initiative

The GSMA Future Networks programme will focus on how operators can widen the reach of their 5G networks and maximise the use of their edge assets by utilising cloud technology principles. The programme’s first initiative in this area, Operator Platform, will engage operators who want to utilise their physical network assets in order to enable 3^rd party developers to deploy enterprise applications on Multi-access Edge Computing (MEC) with interoperable reach at localised points of presence and across multiple operators. This topic will be the focus for the GSMA Operator Platform Seminar during MWC2020, as well as a series of 5G Developer Labs held globally throughout the rest of the year.