Visiting the Edge Computing World conference last week, I observed a number of interesting aspects that I would like to share here. First of all, the reason for my title is my observation that technical terms such as Edge or Cloud are used rather freely these days, almost causing the terms to become meaningless. Almost any aggregation node in a distributed architecture was called a “Micro Cloud” or “Cloud at the Edge.”
In doing so, the ignorance of some key characteristics of the cloud apparently mattered little; scalability or elasticity first and foremost. From the perspective of the edge node, the aggregation node does not offer any characteristic cloud functions (the aggregation node is basically a black box, an aggregation and processing unit without any other properties). From the point of view of nodes located higher up in the hierarchy, the aggregation node also does not offer any characteristic cloud functions. Of course, additional edge nodes cannot simply be created; after all, we are usually talking about hardware here and not about virtual capacities such as additional virtual machines.
It makes little sense to reduce Edge and Cloud to such a small subset of their properties, as both terms then degenerate into empty shells rather than meaningful concepts. If one does so, however, the question arises as to whether Edge is already in the process of dissolution, or whether we need to redefine terms (although there is actually little reason for this in purely factual terms) to reflect the changed view of the various layers within the Edge-Cloud continuum. This raises the suspicion that business-motivated forces are at play.
Edge Interoperability? Or Vendor Lock-In?
This brings me to another aspect: interoperability. With traditional service providers now selling hardware (e.g., Amazon Snow) and traditional hardware providers selling services (e.g., Schneider Electric), the number of players on both sides increases. This bears the risk of reduced interoperability, as the new full spectrum players sell the advantages of “homogeneous solutions” to their customers. As long as one remains within a system, there will certainly be advantages, since hardware and software are coordinated with each other and can fall back on system-specific exchange mechanisms. However, vendor lock-in is inevitable when vendors define their own exchange formats, interfaces, and conceptual models. As soon as different systems have to be integrated, the development of corresponding bridges or transformations will be unavoidable.
With the growing number of systems and corresponding platforms (which are two other terms that are frequently used interchangeably), the number of platform-specific formats and interfaces increases – and interoperability suffers. What has been achieved in other domains, such as Earth Observation – where agreements on standardized interfaces and data models have boosted interoperability – is still somewhat new to the Edge community. The Edge community is in the very early stages of moving toward interoperable (perhaps even open) systems that significantly simplify the generation of complex workflows across system or platform boundaries – or enable them in the first place. Going beyond individual systems is unavoidable: multi-system workflows enable deep insights into domain-specific systems or environments and are necessary to holistically address the grand challenges of our century, such as our changing climate.
Sustainable solutions for the greater good
There is currently still a lot of money to be made with custom platforms. It remains to be seen to what extent these platforms will be suitable for addressing the major challenges of our century. Edge is in a gold-rush mood and I don’t begrudge anyone developing business successes from it. However, the world is extremely complex and I doubt that this complexity can be sufficiently taken into account with the current systems trimmed for depreciation. I heard about examples of fish farms running fourteen parallel Edge systems to monitor the status of the farm. This is fourteen parallel dashboards. Other organizations report that they maintain over 100 software solutions to monitor the health status of manufacturing machines. Most of these are not interoperable, which results in additional costs as soon as several machines form a unit that needs to be monitored as a system. So, where do we stand with Edge? With over 270 trillion USD projected revenue over the next 30 years (numbers McKinsey reported at the conference), climate change alone will produce many new unicorns. Let’s hope that interoperability doesn’t fall by the wayside – or that a sufficient number of these unicorns make their profit with sustainable solutions that contribute to the greater good.