DESIGN CONTEXT
Evaluate the integration work
Use a first project to test whether shared machine data makes an operational task easier.
Map the current task
Choose one question, such as shift utilisation. Record where the required data lives, who extracts it, how identifiers are matched, and how the answer is checked. Existing systems may already share some of this data; build on what works.
Define the change you expect
A shared namespace can let several applications use the same machine signals. The intended benefit is less repeated mapping and collection work. It requires agreed data contracts, reliable source connections, and a way to identify missing or stale messages.
For the fn-uns example, the pipeline stores state durations and production records before the KPI API queries them. Follow the architecture to see those responsibilities.
Measure the pilot
| Measure | What to compare |
|---|---|
| Time to answer | Time from a question to a checked result, including manual preparation. |
| Data coverage | Expected messages and records compared with what actually arrived. |
| Freshness | Age of a value when it is used, including collection and processing delay. |
| Agreement | Difference between calculated metrics and observed production or an agreed reference. |
| Ongoing effort | Work needed to operate, change, and recover the new path. |
Review with the people using the result
Operations checks that states and shift boundaries reflect the floor. Engineering checks signal quality and failure behavior. The software owner checks service health and recovery. Together, decide whether to expand the pilot.