Liberum Arbitrium, or if you prefer free will, is a philosophical concept according to which you have the possibility to choose what to think and how to act solely based on your will to do so. Free will, is the idealisation of a human condition, its degree depends on the awareness of the forces that influence and constrain us along to the structural possibility of carrying out our choices.
Let’s now consider architects and developers of Consumer and Industrial Internet of Things (IoT) applications — I would argue that cloud-centricity has constrained their Architectural Liberum Arbitrium.
Why so? Cloud-centric technologies are mature, operationally convenient and heavily promoted by tech giants and their ecosystem. Yet, they have both influenced and constrained the choices of architects and developers. On one hand the technical community has been influenced to believe that this is the way to go; on another hand the alternatives have been limited and somewhat constraining.
Luckily, there is a growing awareness of the constraints imposed by Cloud-centric architectures as well as a better understanding of the assumptions that, when violated, make cloud architectures impracticable. More precisely, the viability of a cloud centric architectures relies on all of the assumptions listed below to hold true:
- Connectivity. Devices are always (sufficiently often) connected.
- Latency. Applications can tolerate the latency associated with pushing data from devices to the cloud, processing information on the cloud and eventually sending back some control information.
- Throughput. There is sufficient bandwidth to push data to the cloud.
- Cost of Connectivity. The cost of connectivity is negligible.
- Security. It is fine to give away our data.
In reality, these assumptions are violated by virtually all Industrial Internet of Things applications, where notable examples are provided by application domains such as Smart Factories, Robotics Applications, Industrial/Agricultural/Consumer Autonomous Vehicles, and Smart Grids. That said, even in Consumer IoT applications for which connectivity, latency, throughput and cost of connectivity are not an issue, Micro-architectural Data Sampling (MDS) vulnerabilities exposed through the RIDL and Fallout attacks should be extremely concerning for data privacy — thus raising concerns with respect to the viability of cloud-centric architectures.
These challenges posed by Cloud-centric architectures have been actively debated by a niche of technologist through the past few years. The increasing awareness of the architectural limitations imposed by this paradigm, along with the security concerns, has motivated the emergence of Edge and Fog Computing as a decentralised/distributed alternative to the cloud.
What we have missed thus far, is an ecosystem that could accelerate the development and the convergence around an Open Technology Ecosystem for Fog and Edge Computing.
The Eclipse Edge NativeWG has gathered a world-class working group around the shared objective of accelerating the development of Open and Innovative Technologies that will enable edge and fog-computing architectures, thus addressing the limitations imposed by cloud-centric architectures.
The Eclipse Edge Native WG charter, is quite clear on its intent:
To deliver production-ready platforms for the development, operation, and management of edge native applications deployed to heterogeneous environments where computational power and data storage are physically distributed wherever they are needed.
In other terms, the goal of the Eclipse Edge Native WG is to provide a viable alternative to cloud-centric architectures to those applications and users that want to be able to process data closer to where it is produced, maintain control over their data and have the free will to decide where to run what in the continuum that goes from the data-centre to the thing.
In a way, I’d say that the Eclipse Edge Native WG is here to give you back the well deserved Architectural Liberum Arbitrium.