EMBEDDED TECHNOLOGY
Working at the edge demands the right infrastructure choices
By Iwona Zalewska, regional director for UK & Ireland, DRAM Business Manager, EMEA Region, Kingston Technology
E
mbedded systems, whether for industrial automation, offshore IT, or for specialist applications such as surveillance, are being increasingly deployed as an economical and efficient way to enhance operations. As computing components become smaller, more powerful and more connected, they have been incorporated into machines and systems to automate tasks.
Nowhere have these developments been felt more acutely than in remote or inaccessible locations where temperatures can fluctuate, vibrations and shocks are common, and dust and humidity are part of the territory, along with an unstable power supply. To underpin the connectivity of industrial systems, computing has moved closer to the edge, processing data where it is generated, rather than relying on distant cloud servers. This reduces latency and bandwidth requirements, but can make it challenging to support regular patching, component replacement or on-site engineering. Which is why hardware component selection for embedded solutions is critical. Consumer-grade data storage or memory solutions, for example, will not cut it in an industrial environment where predictability and long-term availability are prioritised over benchmark performance or price. Typical risks include unexpected end-of-life, changes to the underlying bill of materials, less predictable behaviour during power interruption and fewer mechanisms for detecting and recovering from data errors.
Selecting the wrong component introduces risk. Industrial-grade solutions, including data storage, approach the challenge differently. Rather than focusing only on peak throughput, they are designed to deliver consistent performance and data integrity over time. Technologies such as advanced error-correcting code (ECC), bad block management, power-loss protection and automatic refresh help to maintain data integrity even as storage devices age and operating conditions change. For engineers with responsibility for the smooth operation of these systems, storage that remains predictable throughout the expected operating lifecycle of the system will always
40
win over storage that is simply speedier or cheaper.
Lifecycle planning starts with design
Flash storage (SSDs, USB drives and memory cards) has finite write endurance, making forecasting an essential part of system design. Embedded applications often have reasonably predictable workloads, allowing engineers to model Total Bytes Written (TBW), assess write amplification and optimise workloads to extend component life. Health monitoring can provide another layer of protection, including the use of JEDEC- defined lifetime indicators to identify declining device health before failure occurs. Using the built-in SSD SMART reporting, features such as ‘percentage used’, ‘temperature’ and other values can be monitored.
Industrial systems also require stability beyond individual components. Controlled Bills of Materials (BOMs), stable controller and NAND combinations, and formal Product Change Notifications allow manufacturers to manage component transitions without introducing an unnecessary increase in compliance risk. This is a measurable benefit to companies managing these systems, ensuring engineering continuity throughout a product’s lifecycle.
Whether deployed in an oil and gas pipeline or in surveillance equipment, real resilience starts with taking a whole-system approach to reliability. Every element, from hardware, memory and storage to firmware, thermal
SEPTEMBER 2026 | ELECTRONICS FOR ENGINEERS
management, power delivery and lifecycle planning, must work together as intended. Critical components such as SSDs need to be purpose-built for sustained industrial workloads, while DRAM modules should be engineered and rigorously tested to deliver dependable performance over the long term. A failure or weakness at any point in this chain can ultimately compromise the reliability of the entire platform.
Working at the edge
Real-world deployments reinforce this principle, and a great example is SNUC Systems. The company provides rugged technology for customers whose modern edge workloads like AI inference, real- time analytics and secure data capture are essential to their function. Its systems are deployed in traffic lights, point-of- sale terminals, police vehicles, oil rigs and battlefield forward-operating bases. A single hardware failure could halt critical operations. SNUC Systems works closely with Kingston Technology to ensure that its components deliver the core strength needed for its technology.
For any organisation deploying mission- critical edge systems, reliability coupled with long-term performance can only be engineered through the interaction of hardware, software, environment, workload and lifecycle planning. Infrastructure choices made early in the design process will make the difference between years of uptime or solutions that are hard and expensive to fix.
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50
