Industry 4.0: Operational Technology and Edge Computing

Overcoming Latency in Real-Time Industrial IoT

Manufacturers must use data from multiple machines, processes, and systems to adapt the manufacturing process in real-time manufacturing environments. These processes generate vast amounts of data and need machine learning to determine the best action based on the data analysis.

Collecting, aggregating, and filtering the data to send outcomes to a central server requires powerful edge computing systems, which are critical for AI and ML, rather than in the cloud. For example, latency over 3-5 milliseconds makes it impossible to accurately control manufacturing robotics.

Edge computing systems and networks also provide bidirectional information - data sent and collected at the same time – to deliver uninterrupted inventory updates 50 times faster than a cloud environment and avoid overstocking, stock outages, and shrinkage, particularly during periods of peak demand or when extreme situations occur.

Sending a request to a data center across the globe and waiting for a response to return takes time. As a result, cloud computing isn’t ideal for many mission-critical manufacturing applications. Manufacturing without this low latency can’t experience the full benefits of industrial IoT. If a connected machine on an assembly line recognizes a malfunction, any delay in transmitting that signal could be costly. It could lead to mechanical damage or even employee injury.

Cybersecurity Risks in Edge and Industrial IoT Deployments

Security should concern all smart manufacturers. Digital manufacturing requires agility and security. The continuously increasing number of IoT devices in the network means more potential vulnerabilities are open to cyberattacks. The paths used by bad actors to gain unlawful access to networks may easily penetrate endpoint devices or nodes because they have limited encryption or firewall protections. This highlights the urgent need for stronger cybersecurity in edge computing environments.

The centralized nature of cloud computing makes it vulnerable to many forms of cyberattacks. If a factory distributes processing and storage functions throughout the edge, however, no single attack could bring down the whole network. Since computing happens closer to the data source, less data is at risk at any given moment, strengthening cybersecurity in edge computing scenarios.

As more companies install secure edge networks, the addition of devices capable of performing increasingly complex functions further away from centralized servers will also increase. Addressing security risks, achieving economies of scale, and protecting sensitive data from bad actors are some barriers remaining before edge computing gains widespread acceptance.

Manufacturers also want to know how to sync IT and OT at scale securely in a workplace where staff skilled in both OT and cybersecurity disciplines are scarce.

The OT applications from one or more sensors, IoT devices, robots, or other equipment run as small virtual machines (VMs) on an SC//HyperCore cluster. Securing the Scale Computing cluster ensures all VMs hosting OT applications benefit from centralized protection strategies, reinforcing overall OT security. You no longer have to worry about the protection of individual devices. Scale Computing Platform provides built-in disaster recovery options to protect your data and applications. Even in those worst-case scenarios, we can help make sure you experience minimal downtime, critical for maintaining both operational continuity and secure edge networks.

Modernizing Operational Technology for Data-Driven Decisions

One of the challenges faced by manufacturers is the ability to access data from their machines, processes, and systems.

The factory floor typically includes many proprietary platforms and hardware that don’t communicate with one another. Operational technology is still largely traditional, lacking the standardized protocols of IT systems. This gap is one of the main drivers behind OT/IT convergence projects in industrial environments.

For example, extracting high-frequency data, such as real-time scanning of sheet metal to detect stress fractures, monitoring oven fill levels to maintain quality control, analyzing flow rates through industrial piping, or tracking the cycle times of automated machines, can quickly overwhelm a centralized server.

This is where industrial edge computing becomes essential. Enabling data processing close to the source allows faster and more efficient analytics while reducing the load on core IT infrastructure. In a real-world deployment, a food and beverage manufacturer used industrial edge computing to monitor fill levels in high-speed bottling lines, detecting deviations in milliseconds and adjusting output in real time, without needing to push all data to a cloud server.

IoT platforms now connect sensors, industrial equipment, control systems, and edge-based computing platforms to provide better visibility and control. These sensors and applications generate massive amounts of machine data that must be analyzed immediately for plant-wide visibility and smarter decision-making.

Industrial edge computing enables manufacturers to filter, aggregate, and process this data directly at the edge, reducing the volume sent to central servers, whether on-premises or in the cloud. In another case, an automotive plant used edge analytics to monitor torque readings from robotic arms, identifying tool wear early and scheduling maintenance before costly downtime occurred.

By advancing OT/IT convergence and embedding intelligence directly into operational technology, manufacturers can unlock real-time insights, drive efficiency, and reduce latency across production processes.

Predictive Maintenance Powered by Edge Analytics

Predictive maintenance is not an easy process to implement effectively. It requires strong IT/OT integration, where operational technology and information systems are connected to provide the necessary data for analysis. Detecting when a machine might fail through data analytics and mitigating this by conducting maintenance before a potential breakdown limits unplanned downtime risk. Predicting the outcome of the analytics is hindered by a limitation of measured variables and immature ML.

The effectiveness of predictive maintenance depends on the quality and volume of machine data available, and the ability to analyze it in real time. However, traditional cloud-based analytics can struggle with latency, data overload, and bandwidth costs. This is where edge analytics provides a critical advantage.

Running edge computing systems closer to the source means large amounts of data can be processed locally. This reduces dependency on the cloud and allows for condition-based alerts to be generated instantly. For example, edge analytics can monitor vibration, temperature, or pressure fluctuations in industrial equipment, triggering real-time alerts when readings exceed safe thresholds.

In addition to generating real-time condition-based alerts, edge analytics supports local machine learning models that can detect patterns leading up to equipment failure. While some limitations remain, such as limited measured variables and evolving ML accuracy, the combination of edge analytics and IT/OT integration enables increasingly accurate, real-time predictions without relying on constant cloud connectivity. This approach strengthens predictive maintenance strategies and drives smarter, more proactive equipment management.

Reducing Unplanned Downtime with High Availability Infrastructure

The cost of downtime increases in every industry. As business moves to a 24-hour economy, there are no more maintenance windows available. This is one of the more difficult areas to calculate. Based on the reports of analysts and research organizations, the impact is massive. According to Siemens' The True Cost of Downtime 2024 report, unplanned downtime now costs the world's 500 largest companies approximately $1.4 trillion annually, equating to 11% of their total revenues. In the automotive sector specifically, the cost of an idle production line at a major plant has surged to $2.3 million per hour, more than double the cost reported in 2019.

Due to software updates and upgrades, downtime often takes careful planning as systems won’t be available for a certain period. Continuous operations, whether during a software update or even when a disk or a complete server fails, is critical.

Implementing high availability infrastructure, particularly through edge computing, allows for real-time monitoring and rapid response to potential issues, minimizing disruptions. By processing data closer to the source, edge computing reduces latency and dependency on centralized systems, ensuring continuous operations even during network failures or maintenance activities.

Managing the Complexity of Industrial Edge Deployments

Organizations departing from less expensive, established system architectures will inevitably face some complexities in their edge deployment. Building an industrial edge architecture is costly, involves significant planning and logistical preparations, and requires considerable software and hardware resources. For example, adding more endpoint devices or a new inventory storage area requires more servers and peripheral devices. This also increases the overhead expenses necessary to maintain them.

In contrast, a centralized server utilizes the same infrastructure regardless of changes or additions to the factory layout.

AR/VR Enablement for Smarter Training & Inspections

Adopting AR and VR is another challenge that manufacturers face. However, AR in manufacturing offers highly efficient methods to train staff on health and safety, new production processes, fault detection inspection, and more.

While these experiences can be powerful, running AR/VR applications through the cloud introduces latency that disrupts performance, making it less effective on the factory floor. By contrast, virtual reality edge applications process data and render streams locally on edge devices, eliminating most latency issues. This enables a smoother, more responsive user experience and supports real-time interaction during training or inspections.

Simplifying Industrial Data Analytics at the Edge

Data is the foundation of the Industry 4.0 revolution. IoT devices and sensors collect massive amounts of data, providing a valuable base for improved decision-making through Industry 4.0 data analytics. However, analyzing all this data requires considerable storage, bandwidth, and computing power.

Edge computing overcomes these concerns in two ways. First, edge data processing occurs at or near the source, so the overall process is much faster. Second, each data point within the smart factory processes its own information, so no one system must handle everything. Processing data at the edge reduces the strain on individual systems and enhances the speed and relevance of Industry 4.0 data analytics. Since it’s segmented, it’s easier to sift through and find relevant information.

Enhancing Interoperability Without Protocol Standards

Seamless industrial interoperability between devices and systems is key for every smart manufacturer. At the same time, this is one of the bigger challenges since there is no standard protocol. If these machines could translate protocols, though, they would make themselves more compatible.

Moving compute functions to edge computing systems located close to the machinery, such as sensors and PLC back boxes, helps to address this challenge. Located on an edge computing system, devices can convert signals themselves and work with various systems.

Solutions like SC//HyperCore play a key role in enabling flexible edge environments. By providing a robust and scalable platform, Scale Computing allows manufacturers to run virtualized workloads at the edge, making it easier to manage protocol translation and support diverse industrial systems.

Manufacturing Challenges: Bandwidth, Data Loss & Legacy Systems

Shane Rogers, IT Director of Harrison Steel, had an open mind when looking for technologies that would help to optimize the manufacturing operations and business system processes, improving its competitive abilities.

For years, Harrison Steel collected data. Data points collected from PLC devices have been there for decades. It used a sneakernet to grab these data points, but it never looked at data for other areas like PMs or the quality of the parts. As more engineers and technology got involved, Harrison Steel realized it could use that data to make better quality parts. This was based on human data collection, starting 15-20 years ago. The IOPS of this data collection, however, was extremely low, making automated data collection and real-time analytics impossible. This was a major concern for Harrison Steel.

Working with machines that have “tons” of sensors, Harrison Steel started to look at what more it could do with the data it could collect and what type of networking it would need. Rogers began to realize, after the company coupled the first machines on its network to automate data collection, that it was losing data and overloading the network. It had to stop because there was simply not enough bandwidth.

Harrison Steel runs very modern sensor technology, but also still has robots running, for example, Windows CE. The problem with these older systems is that they cannot connect to anything other than just to dump data, which is still a manual process.

Evaluating Edge Alternatives: From Legacy to SC//HyperCore

Harrison Steel started researching the market for a separate manufacturing production network, such as Rockwell and others, just to collect OT data from the machines and ovens. One of the objections to adopting this was the high cost of expanding or upgrading the network when it would need more bandwidth. As more and more machines and vast amounts of sensory data come in every year, the cost and flexibility became a concern.

Halfway through the research process, Rogers learned that Scale Computing, a vendor he was already familiar with, had introduced the HE100 Series, a new category of edge computing devices. Harrison Steel was running its data center operations on various Scale Computing HyperCore clusters for years and with great satisfaction.

The Breakthrough: Implementing HE150 Edge Clusters

For Rogers, this announcement was a light bulb turning on. He realized he could place little boxes with three-node clusters in a full, highly available setup close to the machines. With machine data stored on the cluster, no network congestion would occur. It made real-time analytics possible. At regular intervals, the cluster would be offloaded over the network back to the data center in a controlled way.

Harrison Steel has implemented the HE150 clusters running SC//HyperCore and Rockwell Automation in the factory. With these clusters, repeated across the factory floor and connected to all types of machinery and robots, extensive data collection and analysis of mold fill time, flow rates, vibration, and temperature data (among others) takes place. This way, Harrison Steel can make better-informed business decisions at greater speed.

Another great advantage of the setup is the seamless integration with the clusters in their data center running core applications like the Electronic Production System (PMS), Reverse Scheduler, ERP, and InforEAM. Although Harrison Steel uses Scale Computing in the data center as well, the Scale Computing edge solutions can also work with other data center systems and infrastructure.

Business Impact: ROI and Cost Savings at the Edge

From an ROI perspective, building a dedicated network for the data collection would have cost Harrison $500,000 to $1M. This is much more than the dozen HE150 edge clusters implemented (ranging from $7-12,000 per 3-node cluster, including the multi-year SC//HyperCore subscriptions). Besides these noticeable savings, the self-healing technology embedded in the SC//HyperCore system minimizes ongoing management hours.

Reducing Storage Costs with Edge-First Architecture

Implementing Industry 4.0 manufacturing processes means large amounts of data require substantial storage. Traditional local storage options are inconvenient, and cloud storage can be expensive. Storing data at the edge is more efficient. The edge becomes the gateway, allowing local data analytics, sending results or summarized data to the cloud, and saving central storage costs. Storing data on edge computing devices enables manufacturers to select what gets sent to the cloud.

The result is a segmented, organized, and affordable data storage solution.

Scale Computing appliances efficiently use integrated hybrid, all-flash, or NVMe storage. Starting as small as 250 GB, a 3-node micro edge cluster for small, highly available OT applications used on the factory floor to edge systems can run 10s of virtual machines.

Centralized Fleet Management for Distributed Edge Sites

With SC//Platform, you can run applications and process data outside centralized data centers, at the edge of your network closest to where that data is created and utilized. Leverage Scale Computing Fleet Manager to centrally monitor and manage hundreds or thousands of distributed edge infrastructure deployments with few or no onsite IT personnel.

SC//HyperCore and SC//Fleet Manager: Platform Overview

SC//Platform consists of two key applications: SC//HyperCore, the hyperconverged IT infrastructure server environment, and SC//Fleet Manager, no matter if you need to manage just a few or 50,000 clusters or locations.

The award-winning SC//HyperCore delivers on-premises edge computing with high availability and disaster recovery to remote locations at an affordable entry-level cost. All servers can be deployed quickly, managed locally or remotely, and can almost instantly self-heal. Benefit from affordable edge computing infrastructure without sacrificing reliability, ease of deployment, and ease of use.

SC//Fleet Manager is the first cloud-hosted monitoring and management tool built for hyperconverged edge computing infrastructure at scale. It’s now easier than ever to securely monitor and manage your entire fleet of clusters running SC//HyperCore. For IT Managers with multiple clusters, Scale Computing's industry-leading low maintenance time will be reduced by 50% or more.

Modernization with Kubernetes, Containers & Low-Code Apps

To modernize operational technologies, manufacturers adopt low-code and container technologies to boost their application efficiency more frequently to modernize operational technologies.

We often hear customers ask if it is possible to run containers and Kubernetes on SC//Platform. The answer is, yes! We fully expect the adoption of containers to increase over the coming years as more and more application vendors take advantage of the provided portability. The more complex these container-based applications become, the more likely they will be deployed with an orchestration tool such as Kubernetes.

Containers. A container bundles together the entire runtime environment, including the application, its dependencies, libraries/binaries, and the configuration files needed to run it. Containers are outstanding in their ability to provide consistency from one environment to another (e.g., from the developer’s test environment to a VM running on SC//Platform).

Kubernetes. Kubernetes is a widely adopted, open-source orchestration tool for managing and running containers in production environments. It provides a framework for containers’ resiliency, networking, load-balancing, and scaling. It is common to see these tools as a part of a larger CI/CD (continuous integration and continuous delivery) pipeline.

Low-code applications. Low-code platforms give you the tools to assemble the right business experience across domains, keeping your digital thread unbroken from strategy to planning to design to manufacturing. Across the value chain, manufacturers see the possibilities for new efficiencies, new connections, and new ways of engagement.