Hana Pre

Overview

The hana module ingests metadata from SAP HANA into DataHub. It targets both SAP HANA Cloud and on-premise HANA tenants and is intended for production ingestion workflows.

The connector extends DataHub's SQLAlchemySource, so the standard SQL extraction path (tables, regular views, schema reflection, profiling, classification, stateful deletion detection, container hierarchy, test-connection) is the same as for other SQLAlchemy-backed sources. On top of that, three HANA-specific paths can be enabled:

• Calculation Views — set include_calculation_views: true to pull activated calculation views from _SYS_REPO.ACTIVE_OBJECT and parse column-level lineage from their XML definitions. Filter the set with calculation_view_pattern (matched against <package_id>.<view_name>); the inherited view_pattern filters regular SQL views and does not apply here.
• Stored Procedures — enabled by default via include_stored_procedures: true. Each procedure becomes a DataJob grouped under a per-schema DataFlow; lineage is parsed from the procedure body.
• Query Usage — set include_query_usage: true to mine _SYS_STATISTICS.HOST_SQL_PLAN_CACHE for observed queries and feed them through the SQL parsing aggregator. Combine with include_usage_stats: true for DatasetUsageStatistics rollups and include_operational_stats: true for read operations.

Deployment compatibility

Most features work on every supported HANA tenant, but calculation-view extraction is on-premise / self-managed only:

If you run include_calculation_views: true against HANA Cloud, the calc-view extractor logs a warning and falls back to a no-op rather than failing the run, so the rest of metadata ingestion completes normally.

Prerequisites

Database driver

The connector uses SAP's official hdbcli driver via sqlalchemy-hana. Both are installed automatically when you install the hana extra.

Architecture note: hdbcli ships precompiled wheels for x86_64 only. On aarch64 / arm64 (Apple Silicon, AWS Graviton) you must run the ingestion process under an x86_64 Python interpreter (e.g. via Rosetta or an x86_64 container).

Network connectivity

The default port for the SAP hdbcli driver is 30015 (single-container instances) or the tenant port surfaced in HANA Cockpit (multitenant database container deployments). Ensure the ingestion process can reach the tenant on that port.

Permissions

Run the following as a HANA administrator (a user with ROLE ADMIN and USER ADMIN) to create a DataHub-specific role and user. Grants are layered by capability — comment out the sections for features you don't intend to enable.

-- 1. Role + user
CREATE ROLE DATAHUB_ROLE;

CREATE USER DATAHUB_USER PASSWORD "<your-strong-password>" NO FORCE_FIRST_PASSWORD_CHANGE;
GRANT DATAHUB_ROLE TO DATAHUB_USER;

-- 2. Baseline metadata: tables, regular views, schema reflection.
--    Repeat for every schema you want to ingest.
GRANT SELECT ON SCHEMA "<YOUR_SCHEMA>" TO DATAHUB_ROLE;

-- 3. Stored procedures (required when `include_stored_procedures: true`,
--    which is the default).
GRANT SELECT ON SYS.PROCEDURES TO DATAHUB_ROLE;
GRANT SELECT ON SYS.PROCEDURE_PARAMETERS TO DATAHUB_ROLE;

-- 4. Calculation views — ON-PREMISE / SELF-MANAGED ONLY.
--    Requires the SAP HANA XS-classic repository (`_SYS_REPO`). Skip this
--    block entirely on SAP HANA Cloud / HDI-only deployments.
GRANT SELECT ON _SYS_REPO.ACTIVE_OBJECT TO DATAHUB_ROLE;
GRANT SELECT ON SYS.VIEW_COLUMNS TO DATAHUB_ROLE;
GRANT SELECT ON SCHEMA _SYS_BIC TO DATAHUB_ROLE;

-- 5. Query usage from `_SYS_STATISTICS.HOST_SQL_PLAN_CACHE`
--    (required when `include_query_usage: true`).
--    `MONITORING` is the lowest-privilege role that covers it; alternatively
--    use the broader `CATALOG READ` system privilege.
GRANT MONITORING TO DATAHUB_ROLE;
-- or, equivalently:
-- GRANT CATALOG READ TO DATAHUB_ROLE;

-- 6. Profiling (only required when `profiling.enabled: true`).
--    Profiling reads sample data from the tables it inspects, so it inherits
--    the SELECT grants from step 2 — no extra privileges needed beyond that.

Capability-to-grant matrix

The MONITORING role is included with HANA's standard role catalog (see SAP note 2147247 for diagnostics if _SYS_STATISTICS looks empty after granting it).