Robust MES Architecture: Hybrid and Edge-Native Approaches

[Cengiz Özemli]

As the use of cloud-based MES (Manufacturing Execution System) becomes widespread in production processes, serious problems such as latency and system freezes are coming to the fore. Edge-native architectures, on the other hand, overcome these problems with local deterministic control while preserving the analytical advantages of the cloud.

Although cloud-based MES systems promised fast and low-cost implementation, latency poses a significant risk for production speed and critical control times. Especially in operations with sensitive cycle times like 20 ms, a 200 ms cloud latency is unacceptable.

### Disadvantages of Cloud-Only MES

• The cost of average IT downtime is around $2.3 million per hour.
• Latency variability (jitter) occurs in multi-tenant cloud systems, leading to an increase in production errors.
• In real-world examples, cloud-induced outages at large companies like Jaguar Land Rover and UNFI halted production and caused millions of dollars in losses.
• Cloud MES contracts include annual price increases of 7-10% and high data egress fees.

### What is Edge-Native MES and How Does It Work?

Edge-native MES keeps critical control processes on the local network, providing fast and reliable production control with <10 ms deterministic latency. Container technologies like K3s, MicroK8s, and WebAssembly modules run on non-virtualized local PCs or DIN-rail gateways. The cloud is used for long-term analytics, KPI tracking, and AI model training.

### Advantages of Edge-Native MES

• Local predictive maintenance models that reduce production line downtimes by 15-25%.
• With <10 ms visual quality feedback, damaged parts are detected early, reducing the risk of rework and recalls.
• Keeping data local narrows the external attack surface and facilitates GDPR/ITAR compliance.

### Four-Question Matrix for MES Architecture Selection

1. Is real-time (<100 ms) needed? If yes, edge is mandatory.
2. Are there data sovereignty restrictions? If yes, data must be stored locally.
3. Financial preferences (Operational expenditure or investment)? 5-year total cost of ownership is evaluated.
4. What are the personnel competencies? DevOps/OT knowledge is required for edge-native.

### Four-Stage Roadmap for MES Transition

• Stage 1: MES functions are cataloged by criticality, latency, and data sensitivity.
• Stage 2: A pilot K3s cluster is set up on a non-critical production line and tested.
• Stage 3: Real-time loops are run in parallel and gradually transitioned to the edge system.
• Stage 4: Isolated clusters are transformed into a hybrid platform managed under GitOps.

This approach allows for escaping cloud lock-ins and increasing operational flexibility without disrupting production.

In conclusion, while cloud systems offer significant advantages for businesses, edge-native hybrid architectures provide more sustainable and robust solutions for production systems like MES where milliseconds are critical.