Integration architecture
Design Microsoft data platforms as part of a heterogeneous enterprise, with clear boundaries between systems.
When should data move, and when should it stay?
Fabric supports interoperability with external platforms through documented connectivity and OneLake access patterns. Choose between movement, supported shortcuts, and mirroring according to freshness, ownership, security, and operational needs.
Choose the integration pattern.
A shortcut points to data; a pipeline moves or transforms it; mirroring has source-specific behavior. These patterns have different prerequisites and operational responsibilities.
- Identify supported sources, formats, authentication methods, and network paths.
- Assess freshness, change handling, egress, and source availability.
- Document what remains authoritative and how changes reach consumers.
Preserve enterprise contracts.
ERP, CRM, operational databases, APIs, and other clouds have their own change cycles and service obligations. Fabric integration must accommodate them rather than treating every source as a passive file store.
- Define schemas, ownership, retention, and schema evolution expectations.
- Review Azure Databricks and existing lake interoperability against supported integration paths.
- Plan recovery when a source, gateway, credential, or downstream consumer is unavailable.
Prove the complete path.
A successful connection test is not an enterprise integration test. Validate business results, permission boundaries, and operations over the path users depend on.
- Reconcile business totals and freshness from source to semantic model.
- Test identity and access under both normal and failure conditions.
- Include observability, deployment ownership, and support handoff in acceptance.
Microsoft documentation
Product guidance checked 17 November 2026. Architecture recommendations are HeadToNet’s editorial analysis; validate feature availability for your tenant, region, and workload.
Put your architecture in context.
Bring your current platform, constraints, and priorities to a Microsoft data-platform assessment discussion.
