Trends in Industrial Edge Computing Development: Architecture Integration, Machine Connectivity, and IT-Driven Operations
#Industry News ·2020-01-09 10:42:41
The field of industrial edge computing is experiencing numerous technological advancements and developments, particularly in terms of machine connectivity, optimization of edge computing performance, and the scalability of Industrial Internet of Things (IIoT) solutions.
Edge computing consists of numerous distributed edge nodes, which are located close to physical data sources. On one hand, these edge nodes are connected to various devices; on the other hand, they are linked to central platforms such as the cloud. Unlike components at the production asset level, edge nodes enable centralized management. Additionally, the collected data can be processed either within the edge nodes or by the central platform.
The edge layer can be examined from multiple dimensions. From an application perspective, it encompasses software applications and their functionalities, such as data preprocessing and data buses. From an infrastructure standpoint, it refers to the deployed information technology (IT) infrastructure, including hardware facilities and operating systems. From an operational perspective, it describes the various tools used to manage and operate the edge layer, such as monitoring tools or tools for handling multi-site software deployment.
The reasons for adopting edge computing in Industrial Internet of Things (IIoT) applications are well-known. Some applications have extremely low latency requirements, which are difficult to maintain through communication with centralized cloud platforms. In certain situations, the volume of data is immense, necessitating extensive preprocessing at the edge layer. Last but not least, some applications are subject to regulatory conditions that may prohibit data from leaving the company's network.
Architectural Trends
Customers looking to build and operate IIoT solutions must consider numerous issues, with one of the most critical being the selection of an appropriate system architecture. Currently, IIoT architectures are showing a trend toward integration, which is partly reflected in the following aspects and characteristics of the edge layer:
Users are deploying cloud platforms but aim to reduce the edge layer's technical dependency on the cloud as much as possible while avoiding vendor lock-in.
Users are dividing the edge layer into two tiers: the lower tier is the factory floor level, and the upper tier is connected to the central platform or cloud, with both tiers being centrally managed.
Users deploy an MQTT broker at the edge layer as a central hub for data traffic. Data is transmitted to the central platform via MQTT or Kafka, while locally running applications can access this MQTT broker.
For IIoT applications, efficient and secure access to machine or device data is critical. Functionally, the requirements for machine connectivity in this context are essentially no different from those of traditional shop-floor applications.
Typical equipment within factories, or equipment that needs to be addressed in retrofit projects (brownfield projects), primarily requires integration of control systems. The collected data must be provided through standard protocols with application-level support, which typically means using OPC Unified Architecture (OPC UA) or Message Queuing Telemetry Transport (MQTT) protocols. Additionally, the ability to effectively handle multiple data sources, such as integrating data or data sources into a single interface, is also crucial.
Upon examining suitable operational models, more pronounced differences between traditional shop-floor applications and IIoT solutions become apparent. In traditional application scenarios, after adding machine connectivity features, these applications can be deployed and operated locally within production facilities. For example, a Human-Machine Interface (HMI), a Supervisory Control and Data Acquisition (SCADA) system, a Manufacturing Execution System (MES) solution, or even a database link for data backup might be configured. Since users of machine connectivity features are often not IT professionals, they require user-friendly IT interfaces.
In contrast, IIoT solutions often require deploying applications or IoT/cloud platforms across multiple production sites. Unlike traditional scenarios, these platforms not only run multiple applications but also continuously evolve throughout the solution lifecycle, driven in part by the short innovation cycles in software and IT. Enterprises deploy dedicated teams to operate solutions, with personnel responsible for multi-site management and possessing extensive IT knowledge. Customers seek to leverage IT-driven operational models and their advantages in solution efficiency and scalability.
Like other solution components, machine connectivity must meet equivalent requirements in flexibility, operational efficiency, and scalability. Today, an increasing number of users no longer view machine connectivity as a production asset but rather as a component of the edge layer. This is due to the numerous significant advantages edge layer components offer in terms of efficiency and scalability.
"In the foreseeable future, software and IT innovation are expected to remain the core drivers of investment in the manufacturing industry. At the same time, the deployment of edge computing in manufacturing will accelerate."
Machine Connectivity
If machine connectivity is to be deployed as part of the edge layer, what requirements must it meet? Here are some key points:
Machine connectivity relies on software modules deployed on standard hardware, managed by end customers in the same way as other edge layer software components. Currently, Docker containers are a common choice for such deployments.
Machine connectivity can be managed using standard IT tools. Generally, this involves popular platforms based on Kubernetes, such as Red Hat OpenShift or SUSE Rancher. In some cases, lighter alternatives like Portainer may also be adopted.
Machine connectivity provides relevant data to common IT monitoring tools such as Prometheus and Grafana.
Machine connectivity offers well-documented and stable interfaces for configuration tasks, using standard protocols suitable for remote, automated, or combined operational modes (such as HTTP REST protocol).
In addition to the technical requirements mentioned above, we are observing growing market interest in using machine connectivity as a service. Users expect flexible pricing models that can be tailored to actual needs (and benefits), eliminating the need for capital expenditure or investment in equipment.
Future Outlook
In the foreseeable future, software and IT innovation are expected to remain the core drivers of investment in the manufacturing industry. Solution architectures are likely to further consolidate around specific standards, with architectural blueprints and best practices providing simpler ways to meet end-user needs. At the same time, the deployment of edge computing in manufacturing will accelerate.
Manufacturing enterprises are increasingly focusing on IT-driven operational models for machine connectivity. Moreover, this trend is expected to strengthen further in the medium to long term, as IT standards and tools gain more acceptance at the shop-floor level and in the OT domain. At that point, the distinction between traditional solutions and IIoT solutions (at least in terms of machine connectivity) will no longer be relevant.
Reprinted from Control Engineering China