Unified Data Engineering with Microsoft Fabric & Azure AI: Bridging OT Telemetry and Enterprise BI

An in-depth guide on unifying high-volume IoT/OT telemetry streams with enterprise ERP data inside Microsoft Fabric OneLake, utilizing DirectLake mode, Delta Parquet storage, and Azure AI models.

Operational telemetry and enterprise data usually live in different worlds — SCADA and IoT streams on one side, ERP and CRM on the other. Microsoft Fabric's OneLake lets you land both in one governed lakehouse and report on them together.

One copy of data with OneLake and DirectLake

Store everything as Delta Parquet in OneLake, and Power BI's DirectLake mode reads it directly — no import refresh, no duplicated extracts. High-frequency telemetry and business data share one source of truth.

• Delta Parquet as the open, single storage format.
• DirectLake for near-real-time BI without import cycles.
• Unified governance and lineage across OT and IT data.

Landing telemetry alongside ERP

(spark.readStream.format("eventhubs").options(**conf).load()
    .select(from_json("body", schema).alias("m"))
    .writeStream.format("delta")
    .option("checkpointLocation", chk)
    .toTable("onelake.telemetry.meter_reads"))

The result is one analytics surface where a plant manager and a finance analyst query the same lakehouse — telemetry, cost, and outcomes side by side.