Data Integration
Data Factory · Integration runtimes · Fabric Data Factory · Event-driven ingestion
Practice questions for this topic →
Data Factory integration runtimes
| IR | Use for |
|---|---|
| Azure IR | Cloud-to-cloud copy, mapping data flows, autoscale, public endpoints |
| Azure IR + managed VNet | Private endpoints to Azure sources managed by Data Factory; runs data flows |
| Self-hosted IR | On-prem or private-network sources, custom drivers; can't run data flows |
| Azure-SSIS IR | Run existing SSIS packages (Data Factory only; Synapse pipelines support only Azure and self-hosted IRs) |
Facts to know
- Pipelines are logical groups of activities that you deploy and schedule as a unit. Three activity groups: data movement, data transformation, control. There's a default soft limit of 120 activities per pipeline.
- Data Factory in Microsoft Fabric is described by Microsoft as the next generation of Azure Data Factory; new data-integration work should start there, and existing factories can upgrade.
- Custom event triggers start processing when an event occurs (for example a blob landing in storage, through Event Grid).
- Pick the tool: ETL/ELT orchestration → Data Factory; code-free transformation at scale → mapping data flows (Azure IR); SQL-based real-time streams → Stream Analytics; Spark engineering and ML → Databricks or Fabric; mirror operational databases into analytics without ETL → Fabric mirroring.
Exam Tips
Tick each tip once you can answer it without looking.
“Copy data from an on-prem SQL Server into Azure on a schedule” → Data Factory + self-hosted integration runtime.
“Run existing SSIS packages in Azure with minimal changes” → Azure-SSIS integration runtime.
“Visual, code-free transformation of large data” → Mapping data flows (Azure IR). Trap: Self-hosted IR can't run data flows.
“Source requires a custom ODBC driver” → Self-hosted IR.
“Process IoT telemetry in real time with SQL-like queries” → Azure Stream Analytics.
“Trigger a pipeline when a file lands in Blob storage” → Storage event trigger (Event Grid).
“Near-real-time replica of Cosmos DB for analytics, new project” → Cosmos DB mirroring in Fabric.
“Pipeline structure” → A pipeline groups activities: data movement, data transformation and control activities. Trap: Default soft limit of 120 activities per pipeline.
“Data Factory in Microsoft Fabric” → Microsoft's next generation of Data Factory; new integration work starts there. Trap: Existing factories can upgrade.
“Integration runtimes Synapse pipelines support” → Azure and self-hosted only. Trap: Azure-SSIS is a Data Factory feature.