Demystifying IoT Edge

There has been a lot of hype and frankly varied understanding of the Internet of Things (IoT) Edge. At one point, IoT Edge was at the peak of the Gartner hype chart. Over the last few years, I have had the opportunity to understand the customer needs and define IoT Edge from a hypothesis to vision to segment creation. And frankly, I have had a lot of scars and lessons learnt too. Best would be to understand the role of IoT Edge, clear some confusion and identify a winning solution. 

IoT can simply be defined as data generated from physical assets used for improving business outcomes. Applications delivering business outcomes have primarily run in the cloud. Over the last decade if there is one word that has caught the technology industry’s attention and growth, it's ‘Cloud’. So, let’s start explaining Edge in the context of cloud. Edge is everything between the source of data coming from a physical asset till the data reaches the cloud. The spectrum between the physical asset and the cloud, called Edge, can be narrow or broad depending on the industry and use case. It could either be a one or multiple hops. 

The common question that comes up is why do we need Edge. Why can't the physical asset not send the data directly to the cloud ? Below are three main reasons why we might need Edge depending on the use case:

1. Bounded by Science : It takes considerable time to send data to the cloud, compute and get back to the asset. Although this turn around time is decreasing, it’s not zero and the use cases where real time action is critical, is only increasing. Added to that, network reliability is not 100%, especially where some of the remote IoT assets reside. Imagine a police vehicle trying to take action in a real time speed chase or consider a water leak in a reservoir, which is in the middle of nowhere, that needs to be plugged to prevent flooding.
2. Bounded by Economics : Sometimes the cost of sending and storing data in the cloud is more expensive than the value of the data itself. Imagine a machine in a manufacturing plant sending temperature readings every second or a parking sensor in a parking spot signaling that the spot is taken. This data might be unnecessary and expensive but yet high fidelity of data is a necessity. 
3. Bounded by Regulations : In many industries and countries, the government and regulations bind what data can be sent to the cloud. 

Let’s try and peel the onion a little more and understand the nuances of Edge. As we discussed above, starting with the concept of cloud and explaining Edge in relation to it, would be much easier. Cloud and Edge have Hardware and Software pieces of technology to it. In the context of cloud, I am sure you have heard of terminologies such as “Server Farms”, “Instances (virtual computing environment)”, “Elasticity”, “cloud native” etc. 

Terminologies such as “Server farms”, “Instances” refer to the Hardware aspects of the cloud and if this doesn't make sense, - I’m hoping this picture can help: 

What this image represents is almost an infinite compute capacity and ability to grow compute as needed based on usage. This is one of the fundamental premises of the cloud that is helpful to keep in mind as we dive into the Edge. 

Edge by definition is constrained, though the level of constraint might vary with use case but the fundamental concept is the fact that things like storage, compute etc. are constrained. It cannot be added and removed dynamically. Once this basic concept is internalized, explanation of Edge becomes fairly easier to follow. 

From a software point of view, the industry and technologists want to follow the latest and greatest advances we have seen in the cloud. For example, having software be micro services based - containerized with good abstraction of functionalities using API’s. However, software on the Edge has to be cognizant of the underlying hardware and it’s limitations. There is a level of optimization needed in Edge. A software module cannot just assume it can spin up more instances/compute as needed. “Code footprint” is a common question on the Edge that you rarely ask in the cloud. 

Besides hardware and software as technologies, deployment of Edge is unique and has its own challenges compared to cloud. Edge is the first place where IT (Information Technology) and OT (Operational Technology) really meet. Edge has to be IT ready and OT friendly. The hardware, infrastructure, workload management and security should be IT ready but the workload itself needs to be OT friendly, with ease of use and seamless synchronization with the cloud.

With this basic understanding of how an Edge differs from cloud and the need for Edge, many organizations and companies have been investing time and effort trying to build the best of technologies to be adapted at the Edge - from workload management migrating from docker/kubernetes in cloud to docker/kubernetes ‘for’ Edge, abstracting the complexities of the Edge so that the workload development stays similar between cloud and Edge, and finally the workloads itself being smart to support just the right amount of functionality based on the compute available and always be in synchronous with the cloud. 

The workload(s) in IoT are typically performing data extraction, data modeling (adding metadata to the data), data brokering (pub/sub), data storage, data visualization, data analysis and finally manual and autonomous automation. The biggest difference across industries is how much of these workload functionalities run in cloud, Edge or a seamless combination of both.

I have seen the need for Edge use cases across every market segment possible - Consumer, Manufacturing, Transportation, Smart City, Utilities, Mining - and the list goes on. Here are some examples about IoT Edge Use Cases to get the creativity juices flowing, 

• A traffic intersection signal changing light color based on pedestrians crossing the street 
• A snow piler dispensing salt on the road based on amount of snow
• A connected car or fleet of trucks/buses/vessels with multiple engine subsystems generating 1000’s of data points every second for real time maneuvering or vehicle operations
• A water pipe transmitting drinking water across the country needing to be shut off when there is a water leak
• A gas valve in a mining field needing to be shut if there is a gas leak
• <and the list goes on…..>

The use cases where IoT Edge is applicable is expanding every day as new use cases emerge for IoT. The trick is in knowing how to make it work seamlessly with the cloud so that the discussion is not about cloud vs Edge but more a spectrum of computing environment optimized for certain use cases but completely abstracted for the application.

The complexity and choice to run the application/logic in Edge or cloud or a combination of both should be completely abstracted and orchestrated by the underlying technology/platform and designed for simple user experience. That’s what makes the platform smart, tough and frankly a winner in this space.  

Author

Anoop Mohan

[email protected]

Note:

Cover Image: Courtesy – Akamai, poftut

This article represents thoughts and opinion of the author.