Concept card

Microsoft Fabric & SAP Integration

As of 2026-07-24T14:00:00Z

Edited by Cédric-Olivier Mâry, sole editor and publication director — 27 years in the SAP ecosystem. How these pages are produced, and where automation is used.

What is Microsoft Fabric & SAP Integration?

Four SAP-to-Fabric integration patterns work (Mirroring, Data Factory connectors, Delta Sharing, OData); two that teams keep trying — direct ODBC and Fabric-to-Datasphere reverse mirroring — are unsupported or simply not GA in 2026.

What it is

Microsoft Fabric is Microsoft's unified analytics platform, bundling OneLake as the common storage layer with Data Factory, Synapse Engineering, Synapse Data Warehousing, Data Science, Real-Time Analytics, and Power BI under a single SaaS subscription. For an SAP analytics consultant, the part that actually matters day to day is narrower than the whole platform: the specific integration surface between SAP and Fabric — four patterns that genuinely work in production, and two patterns Fabric users keep attempting that do not.

Why it matters

Fabric adoption inside enterprises that already run SAP is now common enough that "how do we get SAP data into Fabric without breaking governance" is a recurring architecture question rather than an edge case. Getting the pattern choice wrong is expensive twice over — once in engineering effort spent building a pipeline that will not scale, and again in Fabric Capacity Unit spend that nobody modeled up front.

How it works — the four patterns that work

Why it matters in practice

Delta Sharing has zero Fabric-side Capacity Unit cost since data stays in Datasphere, making it the lowest-TCO pattern for read-heavy, infrequent refreshes.
Mirroring's CU consumption is metered on the Fabric side — a BW/4HANA InfoProvider mirrored every 5 minutes at production volume can eat a disproportionate share of an F64 capacity.
Reverse mirroring (Fabric writing back into Datasphere) is only an announced direction, not a shippable 2026 pattern — don't architect around it yet.

Key points

Mirroring — SAP HANA Cloud / BW/4HANA / Datasphere objects replicated into OneLake as Delta tables; low-latency CDC; no ETL pipeline to maintain.
Data Factory SAP connectors — SAP Table, SAP BW Open Hub, SAP HANA, SAP CDC for batch ingestion into Fabric Lakehouse.
Open Delta Sharing — SAP Datasphere publishes a share; Fabric reads as a OneLake shortcut; DAC enforced at share boundary; lowest Fabric CU cost (data stays in Datasphere).
OData consumption — Fabric Dataflow Gen2 calls SAP OData services; low-volume operational integration; not suitable for analytics-scale volumes.
Anti-pattern 1: direct ODBC from Fabric notebook to SAP HANA — works technically, zero governance and lineage, no Fabric audit trail, unsupported in production.
Anti-pattern 2: reverse mirror Fabric → Datasphere — announced 2024, not GA in 2026; do not promise this to a client as a 2026 deliverable.
CU cost trap: high-frequency Mirroring at production BW/4HANA InfoProvider volume can consume a disproportionate share of an F64 SKU — measure with representative data before committing.
Delta Sharing is the lowest-TCO pattern for read-heavy, infrequently-refreshed datasets: no Fabric CU consumption, governed access, data remains in Datasphere.
Pattern selection decision: Mirroring for low-latency Power BI · Data Factory for batch ETL · Delta Sharing for multi-platform governed exchange at low TCO.
The pattern choice drives both the integration architecture AND the cost model — confirm the Fabric SKU and CU budget with the customer's Azure team before finalising the design.

Terms used on this page

OneLake: Microsoft Fabric's unified data lake — every Fabric tenant has one OneLake; all Fabric workloads (Lakehouse, Warehouse, Power BI, Data Science) read and write the same lake under one access model.
Mirroring: The Fabric pattern of low-latency replication from a source database into OneLake as Delta tables, without operating an ETL pipeline; SAP HANA Cloud, BW/4HANA, and Datasphere mirroring announced 2024 with progressive 2025 GA.
OneLake shortcut: A pointer from one OneLake location to another, or to an external source (S3, ADLS, Delta Sharing); lets Fabric workloads read data in place without copying it — the mechanism behind Delta Sharing integration with SAP Datasphere.
Fabric Capacity (F-SKU): The Microsoft Fabric pricing model based on Capacity Units (CU); SAP-to-Fabric synchronisation via Mirroring at high frequency consumes substantial CU and must be sized against the F-SKU before the architecture is finalised.
Delta Sharing: An open protocol (Linux Foundation) for sharing live data across platforms without copying it; SAP Datasphere can publish Delta Shares that Fabric reads as OneLake shortcuts — the lowest-TCO SAP-to-Fabric pattern in 2026.
SAP BW Open Hub connector: A Fabric Data Factory connector that reads from a BW/4HANA Open Hub Destination (OHD), enabling batch extraction of InfoProvider contents into a Fabric Lakehouse without direct database access.
Data Access Control (DAC): SAP Datasphere's row- and object-level security layer. When Datasphere publishes a Delta Share, DACs are enforced at the share boundary — Fabric consumers see only the rows their Delta Sharing bearer token is authorised for.
SAP CDC connector: A Fabric Data Factory connector that implements Change Data Capture (CDC) against SAP HANA Cloud or S/4HANA tables, enabling near-real-time incremental loads into Fabric Lakehouse without full-table scans.

Sources

Full card available to members. What the full card adds: the full decision framework · the SAP vs Snowflake / Databricks / Fabric comparison · the common pitfalls and their fix · the cheat sheet · the architecture schemas · the code blocks · the facts worth quoting.

Open in the app →