Energizing the Manufacturing Industry With Kubernetes And Cloud Native

The presentation discusses the use of semantic web modeling and shapes constraint language to create a platform-independent description of a factory's environment and enable collaboration on use cases. The system must be scalable, low-latency, and declarative.

• The project uses semantic web modeling and shapes constraint language to create a platform-independent description of a factory's environment and enable collaboration on use cases
• The system must be scalable, low-latency, and declarative
• The system uses Apache Kafka and Apache Flink for data processing and storage
• The time series data is stored in Apache Cassandra

The presentation gives an example of a factory that uses metal sheet cutting machines. The machines require air filters to be on when cutting metal sheets, but the current safety regulation demands that the filter runs on maximum even when the cutter is off, which is not very economical. By using semantic web modeling, the factory can reduce the filter strength and power by retaining the air in the factory without blowing it out and manage the hazard status of the filter coverage to make it cheaper to dispose of. However, there are challenges in changing such systems, such as the need for safety certification and the lack of a formal way for domain experts to talk about their use cases and show the benefits of the system.

We present the Cloud Native Digital Twin architecture of IndustryFusion Foundation (IFF). The goal of IFF is to develop an open-source platform to digitize factory processes and achieve carbon neutrality in metal processing. From the start, IFF chose Cloud Native and Kubernetes as the foundation of all workload management. The machine gateways, the factory servers, and cloud deployments all run on CNCF certified K8s clusters and are, therefore, independent of a specific Cloud or Edge Service Provider. The Digital Twin architecture is fully declarative, based on Semantic Web frameworks like RDF and JSON-LD. We developed a special K8s operator to monitor and deploy the factory processes as Apache Flink jobs, defined with Streaming SQL. We will also describe our experience running K8s and Cloud Native in a manufacturing environment, especially the challenges we overcame with the machine builders and factory owners to make them comfortable with the Cloud Native and K8s approach.