Synapse Pipelines vs Azure Data Factory: when to use each

Azure Synapse Analytics includes a pipelines service that is technically identical to Azure Data Factory, but integrated into the Synapse workspace. The question "ADF or Synapse Pipelines?" has no single answer — it depends on the architecture.

Direct comparison

Feature                     ADF                 Synapse Pipelines
─────────────────────────────────────────────────────────────────
Standalone service          ✓ (standalone)      ✗ (requires Synapse WS)
Integration Runtimes        ✓ Azure/Self-hosted  ✓ Azure (limited)
ADLS trigger                ✓                   ✓
Event Grid trigger          ✓                   ✓
Call Databricks notebook    ✓ (connector)        ✗ (indirect)
Call Synapse notebook       Possible (HTTP)      ✓ (native)
Access Synapse SQL Pool     ✓ (via LS)           ✓ (native)
Managed VNET                ✓                   ✓
Monitoring                  Monitor + ADF UI     Synapse Studio
Separate cost               ✓                   Included in Synapse

When to use Synapse Pipelines

• The whole data stack lives inside the Synapse workspace.
• Orchestration of Spark notebooks and Synapse SQL Pools.
• Teams that prefer a single tool (Synapse Studio).
• Cost: Synapse pipelines have no additional execution cost.

When to use ADF

• Multi-service architecture (Databricks + Synapse + others).
• Self-hosted Integration Runtime for on-premises sources.
• Specific connectors not available in Synapse Pipelines.
• An experienced team with existing ADF pipelines.

Hybrid pattern

ADF → global orchestration and ingestion of external sources
      ↓
Synapse Pipelines → transformations internal to the workspace
      ↓
Synapse SQL/Spark → processing execution

Conclusion

For new data-first projects in Synapse, start with Synapse Pipelines — they are free and integrate natively. If you need a Self-hosted IR or to orchestrate multiple services outside Synapse, add ADF as an external orchestration layer.