Software needs data… and today that means software deployment on desktops, on mobile devices (smartphones, tablets and wearables) and also out the computing ‘edge‘, spanning the expanses of the Internet of Things (IoT) and its many sensors, gauges and other embedded computing units.
On a mission to work out what happens across the various different tiers of edge computing as they traverse our newly interconnected data landscape, we spoke to ThoughtSpot senior analytics evangelist Sonny Rivera to download some much-needed insight.
Rivera is a data analytics specialist and Snowflake Data SuperHero. He uses his experiences to enable customers to create new technology solutions while he also works to provide thought leadership and drive product innovation.
The ‘modern’ data stack
As an early adopter of cloud data analytics platforms and a hands-on data leader with deep experience in data engineering, Rivera spends his life working with advanced analytics and the modern data stack. Most recently, he has focussed on modernising and migrating data platforms to the cloud.
Tasked with defining and describing how the shape of edge computing differs in different deployment styles and environments, Rivera has some solid opinions.
“When it comes to describing the difference and navigating between micro-edge, mini-edge, medium-edge, heavy-edge and multi-access devices for edge computing, I think of it much the same way I consider working with other technologies i.e. identify the key drivers for the most common use cases and requirements in order to be able to form the best strategic development roadmap possible,” said Rivera.
So, for edge computing devices, he urges us to consider factors such as processing power and storage capacity.
“Different edge devices have different processing power and storage capacity, which will affect their ability to handle different types of data and workloads. For example, micro-edge devices, like the temperature sensor on my BBQ smoker, may be sufficient for simple data collection and transmission. In comparison, heavy-edge devices may be required for more power, complex data processing, data transmission and storage tasks, like in the case of [American fast food chicken restaurant] Chick-fil-A,” explained ThoughtSpot’s Rivera.
He also suggests (well, insists, to be honest) that we consider the connectivity options available on edge devices. For example, some devices may only support wired connections, while others support multiple transmission methods, including Bluetooth, WiFi and 5G wireless connections.
“In addition, consider the type of network to which the device connects to, low-power and wide-area networks (LPWAN) or multi-access edge computing (MEC). Often devices may need to connect and transfer from one network to another seamlessly,” specified Rivera.
Power requirements are another significant factor of edge devices, so he says we need to be aware of how they will be powered in the field. The possibilities range from batteries, renewables, AC power, or to a combination.
“Edge devices may be deployed in various environments, from indoor to outdoor, from mild temperate to extreme conditions, or dry and wet conditions. So then, developers will need to consider just how rugged the devices need to be for the use case in hand,” added Rivera.
Security is always a key consideration in any digital product so consider the management and security features offered by the edge devices, such as remote management, over-the-air updates and security features like encryption and authentication.
He also points to cost and says we need to consider the cost of the edge devices, including the upfront cost, as well as the ongoing costs associated with deploying and maintaining the devices. Costs will be heavily driven by the factors listed above.
“Without a clear industry standard (or with our shifting definitions), we can use these factors to categorise edge computing devices,” clarified ThoughtSpot’s Rivera. “The more capabilities needed from each factor, the ‘heavier’ the device gets. Think of it as a multi-dimensional chart. At the lower end of the spectrum, we have micro-edge that utilises just a few capabilities and thus is suitable for less taxing use cases. In contrast, the multi-access use cases require the most capabilities and complexity.”
Total stack topography
Developing for and deploying upon the new world of the emerging edge is not in some way ‘simpler’ because of the scaled-down small form factor of some edge compute zones – with microcontrollers and printed circuit boards (PCBs) in some cases replacing what would be a full computer, server or cloud instance in larger environments – getting it right and doing it well appears to depend not just on what the use case, but also a host of other factors that all go together to create the total stack topography and the ecosystems it exists within.
The computing edge is getting sharper, let’s take care.