One product gets your SAP data out, compliantly, and in real time.
Simplement Roundhouse™ delivers a live, analytics-ready copy of your SAP data to the warehouse, lake, or AI agent of your choice, without the unsupported interfaces that break at the next audit.
Simplement Roundhouse™: built for SAP, ready for everything downstream
Getting data out of SAP has always meant choosing among eight imperfect methods, each trading off compliance, latency, and engineering effort. Roundhouse™ replaces that decision: it extracts through SAP's supported application layer, streams changes in real time, and lands clean, analytics-ready data wherever you need it: Snowflake, Databricks, BigQuery, Azure Synapse, Redshift, Kafka, or a Fabric lakehouse.
Compliant by design
Reads SAP's application layer, not ODP/RFC. No interface that SAP can deprecate out from under you.
Real-time CDC
Change data capture streams inserts, updates, and deletes as they happen, not nightly batch.
Decoded, not raw
Resolves SAP's pool/cluster tables, data types, and deltas into business-ready columns.
Any target
One pipeline to your warehouse, lake, streaming bus, or AI layer, on ECC or S/4HANA.
As a technology partner in each ecosystem, Roundhouse™ lands SAP data natively in your platform of choice. See it in depth on the SAP to Snowflake, SAP to Databricks, and SAP to Microsoft Fabric pages.
Live SAP data your AI agents can actually trust
Roundhouse™ lands current, governed SAP data in your own platform, so AI agents and copilots query the real thing instead of stale extracts. Point an agent at live finance, supply chain, or order data, built on a foundation that stays compliant as SAP evolves.
SAP Note 3255746: keeping your Snowflake, Databricks, and Fabric pipelines running
SAP has restricted third-party tools from using Operational Data Provisioning (ODP) APIs and unauthorized RFC calls for external extraction. If your pipeline to Snowflake, Databricks, or Microsoft Fabric pulls data from ECC, S/4HANA, or RISE through an ODP-based connector, that path is the one under scrutiny, and it's the reason many teams are re-architecting now.
Roundhouse™ extracts through SAP's supported application layer, not the ODP/ODQ frameworks or background RFC table readers that Note 3255746 addresses. Because extraction runs inside the application layer, SAP's own security profiles, authorizations, and business logic stay intact, and there's no ODP interface for a future SAP change to remove from under you.
A change posts in SAP
A transaction updates a table such as BSEG or VBAK in ECC or S/4HANA.
Captured in-app
Roundhouse™, installed via SAINT, captures the delta at the application layer, without ODP.
Streamed to target
The change is streamed securely to your destination as it happens, no intermediate staging DB required.
Analytics-ready
Decoded, typed data lands in Snowflake, Databricks, or Fabric, ready for dbt, Snowpark, or Power BI.
The 8 extraction methods, and how Roundhouse™ handles each
Here's the full landscape of ways to get data out of SAP, rated on a 5-star compliance scale. For each, you'll see a real SAP example and how Roundhouse™ either replaces it or does the same job without the risk.
ODP / ODQ via RFC
Operational Data Provisioning over RFC exposes extractors and CDS-based ODP sources to outside tools. SAP has been explicit that using ODP/ODQ for non-SAP extraction is not licensed or supported, leaving these pipelines one upgrade or audit away from failure.
2LIS_11_VAITM sales-order extractor (or a 0FI_GL_*
finance DataSource) through the RODPS_REPL_* / ODQ RFC function modules. Common in older
Theobald, SLT-to-ODP, and generic-connector setups.SAP Datasphere
Datasphere (formerly Data Warehouse Cloud) federates and replicates SAP data inside SAP BTP. It's governed by design, but adds Capacity Unit costs, keeps semantics in SAP's model, and limits how freely data moves to non-SAP targets.
I_SalesDocumentItem CDS views into a Datasphere space via the
Replication Flow / Premium Outbound Integration, then trying to push that out to Snowflake.Application-Layer Extraction with Simplement Roundhouse™
Roundhouse™ runs inside SAP at the application layer, where it sees data exactly as SAP intends it: decoded, typed, and authorization-aware. It captures changes in real time and streams them to your target with no ODP dependency to break.
BSEG/BKPF (finance documents), VBAK/VBAP
(sales), MSEG (material movements), and even cluster tables like RFBLG, fully
decoded, into Snowflake or Kafka as they post in ECC or S/4HANA.CDS Views
Core Data Services views expose business-meaningful, S/4HANA-aligned data and are a supported, future-facing interface. The catch is coverage and consumption: not every object is released for external use, and extraction still needs an interface (OData, RFC, or ODP) to pull it.
C_SalesDocumentItemDEX or an I_* interface view via OData in
S/4HANA. Clean semantics, but bounded by what's been released and by API throughput limits.Database Log Reading
Reading the database redo/transaction log (e.g. trigger-based or log-based CDC straight off HANA or an underlying DB) captures changes fast, but it sidesteps SAP's application logic, security, and support boundary, and you inherit the job of decoding raw structures yourself.
ACDOCA (the Universal Journal), then having
to reconstruct currency, unit, and pool/cluster logic that the application layer would have applied.ACDOCA arrives correct, not raw.SAP BW / BW4HANA
BW is a known, compliant quantity for traditional analytics, but it adds a full modeling layer (InfoProviders, DTPs, transformations) and is itself often fed by the same extractors now under ODP scrutiny. Justifying it purely to land data in a lake is hard.
0FI_AR_4 accounts-receivable extractor into an aDSO in BW/4HANA, then
exporting via Open Hub to a downstream warehouse. Multiple hops, nightly cadence.Custom ABAP
Hand-written ABAP extractors stay inside supported boundaries and pull exactly what you specify.
The trade is engineering: every object, delta logic, and S/4 conversion (e.g. classic AR/AP tables
folding into ACDOCA) becomes your team's permanent backlog.
LIPS (delivery items) with bespoke delta
pointers. Works, until an upgrade changes the table or someone who wrote it leaves.Generic ETL / iPaaS Tools
General-purpose platforms are convenient and well-integrated downstream, but their SAP connectors frequently call ODP/RFC or BAPIs under the hood, so they quietly inherit Method 01's risk, and rarely decode SAP's trickier structures correctly.
Same SAP source, every modern platform
Roundhouse™ extracts SAP data once, through the supported application layer, and lands it in the
destination your team already runs on. The compliance model doesn't change with the target: the same
BSEG, VBAK, and ACDOCA data streams to Snowflake, Databricks, or
Microsoft Fabric without ODP.
SAP to Databricks Lakehouse
Lakehouse teams want clean, delta-tracked SAP data in Delta Lake without standing up intermediate cloud storage first. Roundhouse™ streams changes from the SAP application layer directly into Delta tables, with the metadata Databricks needs to plug straight into Delta Live Tables and Unity Catalog governance.
BSEG/BKPF (finance) and VBAK/VBAP (sales)
into Delta Lake in real time, decoded and typed, ready for ML feature pipelines, rather than nightly
batch loads into a staging bucket.SAP to Microsoft Fabric OneLake
Azure-centric teams often reach for Azure Data Factory to move SAP data, but ADF's SAP path typically leans on ODP extractors, the same interface under scrutiny in Note 3255746, and adds staging and pipeline overhead. Roundhouse™ streams SAP changes directly into Fabric OneLake in Delta format, skipping the intermediate ADLS staging layer.
ACDOCA (Universal Journal) and MSEG (material movements) into OneLake
as decoded Delta tables, queryable in Power BI Direct Lake mode, without an ADF ODP connector in the
middle.Find out which method you're on, and the compliant way off it
Tell us what you're running today, whether Theobald, SLT, BODS, a homegrown ABAP extractor, anything, and we'll tell you plainly whether you're at risk and what Roundhouse™ would replace it with.