Edge Computing Challenges and Opportunities

Continuing the evolution of IoT, edge computing has been gaining momentum and traction recently.

Personally, I have always advocated that the purpose of IoT is not just to show beautiful charts and colorful screens. It is rather an opportunity to take the data out of edge and use the data to understand the opportunities to bring automation back to edge to improve the overall efficiency of the edge environment.

I have met with over 100+ customers and partners in the past two years. There is a common pattern to the approach taken to leverage IoT. In most cases, there is a fundamental technology strategy that drives most of the IoT projects. The simple technology strategy is, "take the data from edge, land it in cloud and analyze the data". However, when rationalization and maturity sets in, the customers eventually understand that,

• One technology strategy will not be able to support all the use cases and its latency requirements and
• They need to have different analytic infrastructure at different points (environments) to support different use cases and their latency requirements.

So, the fundamental paradigm should be,

"Process the data where it matters when it matters the most".

This will inevitably lead the customers towards embracing "edge computing".

Edge presents an amazing opportunity to transform the business by leveraging real time processing. McKinsey published a list of early edge use cases across different industries and these use cases are only the start of a new evolution.

You can see the use cases here: https://mck.co/3ctTg7O

Embracing edge computing presents these challenges (and opportunities).

• Leveraging edge computing does not remove the need for cloud computing platforms. Cloud technologies (more appropriately, "multi cloud") will be an integral part of future computing platforms and they will continue to serve the needs of "exploratory" analytics along with edge analytics.
• Edge is a loosely defined term with amorphous architectural needs based on the environment. An edge can be a video camera, drone, car, plane, a diesel engine in a ship or the whole ship. So, no one architecture will be sufficient to solve needs of all edge environment.
• Cloud native edge will be key for ease of manageability. Workloads for edge environments (typically AI/ML models) will leverage containers for ease of deployment and continuous model refinement.
• Simplifying interoperability (multiple communication protocols at device level) and external connectivity to cloud or data center (4G, LTE, 5G, WiFi 6) will be key to successful deployment at remote edge environment which typically does not have the luxury of IT support crew.
• Security cannot be a "bolt on" for different environment. Security has to be an inherent and fundamental part of architecture. Leveraging different security solutions for different environment (edge, edge cloud, core, data center and cloud) without consistent management capability will only expose the environment for vulnerabilities.
• Managing all the workloads and service orchestration across varying environment will present a huge challenge and solving the "day 2 or 3" problems in the beginning is key to deriving success out of edge deployments.

The companies which focus on simplifying the edge architecture for "ease" of use case deployment and "continuous" management of workloads will eventually emerge as a leader in a market which has no clear leader emerging as yet.

Finally, I have been an advocate of open standards and open technologies. Edge is at its nascent stage of its evolution. It is important to keep the technology platforms as open as possible without any vendor lock-ins to allow for embracing future technologies which are going to be imminent in eco-system in near future.

In the future articles, I will discuss more about cloud native edge, edge architecture and service orchestration across different environments.

#edge; #edge computing; #IOT; #AI; #ML; #cloud native edge; #IoTsolutions