Industry experts agree that critical infrastructure for internet-of-things (IoT) deployments will continue to shift towards the edge. As new use cases emerge across industries investing in their digital transformation journies, the diversification of edge-enabled physical hardware will continue to expand with technology and business requirements. One of the most popular ways of visualizing the broad offering of physical edge hardware available today is to classify it on a spectrum between thin and thick edge. Lets compare and contrast both ends of that spectrum and then take a look at some of the technical requirements for enterprise IoT deployments for each.
Thin vs. Thick Edge: Which is better?
The short answer is that it depends. An enterprise’s use-case and software capabilities are the primary factors in determining which edge hardware best suits its needs. According to MachNation research of enterprise IoT buyers, roughly 90% of the complexity at the edge is software-related. Thus, while many hardware vendors already provide hardware that adequately supports most IoT solutions, the quality, sophistication, usability, and scalability of vendors’ software and platform for both thin- and thick-edge deployments remain a differentiator. While having the right software to suit enterprise needs is critical, choosing the right edge-hardware can often be the key differentiator between an overly-complex and costly to operate edge deployment and an effective one.
Thin Edge Hardware
MachNation defines a thin edge as an IoT solution supporting and managing decentralised deployments of low-powered, memory and power constrained devices, examples of which include Arm Cortex-M devices, low-power Texas Instruments (TI) devices, Marvell Technology devices, and others.
These devices transmit data northbound with minimal execution of workloads (e.g., filtering, aggregation, or normalisation) or analytics being performed at the site of data origin. Enterprises install thin edge nodes to connect legacy equipment to their cloud solutions, thereby taking advantage of cost savings on equipment.
Thick Edge Hardware
MachNation defines thick edge as an IoT solution supporting and managing high-powered, high-capability edge hardware, examples of which include high-powered IoT gateway devices from companies like Dell and Cisco, Arm Cortex-A devices, x.86-based devices, and others.
These devices can run advanced on-device inference, event processing rules, business logic, analytics, and machine learning algorithms.
Thin and Thick Edge Use Cases
The distinction between a “thin” and “thick” edge use-case boils down to where data is processed and intelligence generated.
Typical thin edge use-cases include:
- Track and trace; battery-powered sensors, continuous transmission of small geo-data packets to the cloud
- Agricultural monitoring; low-powered, transmitting selective low-volume data points to a gateway device
- Smart building; battery-powered sensors that only “wake up” when an action is triggered or when transmitting low-volume data to the centralized monitoring and management platform
In contrast, typical thick edge use-cases include:
- Smart factory; high-memory capacity devices that can store and analyze historical machine-performance data on-device
- Cloud optimization; gateway devices that actively filter out “junk” sensor data before transmitting to the cloud
- Latency-critical operations; high-powered devices capable of running on-device inference in near real-time and executing automatic actions (i.e. powering down a wind-turbine automatically when its rotations-per-minute (RPMs) fall outside acceptible parameters)
Key Decision Factors in Choosing Edge Hardware
Every IoT edge deployment should be purpose-built to fit the use case. Restrictions on available hardware and software can also be determining factors. It is the recommendation of MachNation to have a well-defined hardware and software strategy in place before embarking on an IoT edge journey. Consider these factors when choosing and IoT edge hardware vendor:
- Use-case
- Hardware availability
- Cloud infrastructure to support your edge deployment
- Data requirements around latency and on-device compute resources
For more MachNation research on IoT edge vendors, download the free executive summary of MachNation’s 2022 IoT Edge ScoreCard or contact us to learn more.