Storage & DataFusion
Apache Parquet historian
Telemetry is stored as canonical Parquet partitions under the configured storage root (OPENFDD_STORAGE_URL / workspace historian). Drivers and CSV/package import write columnar batches; DataFusion SQL FDD and Overview analytics scan those files.
Benefits at the edge:
- Efficient columnar scans for trend plots and SQL rules
- No external database server required
- Portable files for backup and offline analysis
Query via REST:
GET/POST /api/historian/queryGET /api/timeseries/readings- Dashboard Plots tab
DataFusion SQL
Fault detection rules are DataFusion SQL statements executed against historian tables. See the Rule Cookbook for full HVAC patterns.
API highlights:
| Method | Path | Purpose |
|---|---|---|
| GET | /api/rules |
List active rules |
| POST | /api/rules/save |
Save rule SQL |
| POST | /api/fdd/run |
Execute FDD run |
| POST | /api/fdd-rules/{id}/activate |
Activate rule |
Rules bind to Haystack point semantics through the assignment graph — configure assignments before expecting meaningful FDD output.
Data management
The Historian storage tab (/data-management) inspects partitions and supports purge-by-source with preview/execute gates. Export CSV before destructive purges.
See DataFusion SQL Rules for examples.