Concept card

SAP Datasphere Catalog (governance layer)

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 SAP Datasphere Catalog (governance layer)?

Datasphere Catalog exists because Joule agents hallucinate when reasoning over uncatalogued data — it gives every object a lineage record, quality score, and named owner so agents can report confidence instead of guessing.

SAP Datasphere Catalog is the metadata and governance layer built into Datasphere that answers a question every large SAP estate eventually has to answer: for any given view, table, or data product, who owns it, how trustworthy is it, who is allowed to see it, and can an AI agent reasoning over it be trusted to say so honestly. It replaced SAP Data Intelligence Cloud, which was retired, and it is billed as part of the Datasphere subscription through Capacity Unit consumption.

What it is and why it matters

Why it matters

It replaced SAP Data Intelligence Cloud, retired 2025-06-30, and bills as Capacity Units inside the Datasphere subscription.
A Tier-1 customer with 5,000+ data products recoups a 6-9 month Catalog enrichment project the first time an agent answers correctly instead of hallucinating.
The Business Glossary is what stops the same KPI being defined three different ways across three Datasphere Spaces.

Key points

Catalog replaced SAP Data Intelligence Cloud (retired 2025-06-30) — any client still referencing DI Cloud for governance needs a migration path to Catalog within the Datasphere subscription.
Joule agent answer quality is directly proportional to Catalog completeness: an agent citing a data product with no owner, no quality score, and no glossary mapping will produce low-confidence answers regardless of model quality.
The Business Glossary is the hardest component to populate — it requires business owners, not IT, to validate term definitions; plan 2–4 weeks of workshop time per major business domain.
Data Access Control layers in Catalog travel with the data product into SAP Analytics Cloud stories and Joule answers, extending the same row/column restrictions that governed the Datasphere Space to every downstream consumer.
Lineage Graph impact analysis is the fastest ROI proof: demonstrate a source-object change and show which downstream stories, models, and Joule prompts would be affected — that single demo typically secures executive sponsorship for the enrichment project.
Catalog is a CU-consuming service — size the CU budget to include Catalog alongside Datasphere compute, Replication Flows, and Analytic Model processing; under-provisioning CUs is the most common cause of Catalog indexing delays.
The Marketplace requires published data products, which requires Spaces to have at least one completed data product with a business description, owner, and quality indicator — a readiness gate before self-service discovery is credible.
For regulated clients (pharma, finance, public sector), Catalog audit trails satisfy the traceability requirements of EU AI Act Art. 12 (record-keeping for high-risk AI) and are the fastest path to demonstrating compliance evidence.

Terms used on this page

Data Product Catalog: The registry of all published data products inside Datasphere Spaces, each with semantic description, tags, classification, quality score, and owner.
Business Glossary: The layer of Datasphere Catalog that maps technical object names to agreed business terms, owners, and definitions — the shared vocabulary that prevents KPI drift across Spaces.
Lineage Graph: The directed acyclic graph in Datasphere Catalog tracing data flows from source connection through transformations to consumption assets; used for impact analysis when source objects change.
Data Access Control (DAC): Row- and column-level restriction layers in Datasphere Catalog that are inherited by data product consumers (SAC stories, Joule agents, Marketplace subscribers) without requiring a separate permission configuration.
Capacity Unit (CU): The billing and resource unit for Datasphere services; Catalog indexing, lineage computation, and quality scoring all consume CUs and must be budgeted alongside Datasphere compute.
Data Marketplace: The self-service discovery surface in Datasphere Catalog where business users browse, request access to, and subscribe to published data products — enabled only after products are catalogued with owner and quality indicator.
Joule grounding: The process by which a Joule agent retrieves Catalog metadata (lineage, owner, quality score, business description) for a data product before generating an answer, converting raw data into a trusted, cited response.
Impact analysis: Datasphere Catalog capability that traces downstream dependencies when a source object changes, showing which data products, stories, and agent prompts will be affected before the change is deployed.

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 →