Integrations · Data warehouses · Available now
Azure Synapse Analytics test data, inside your network
Dedicated and serverless SQL pool connector over TDS (the SQL Server connector's driver and TLS enforcement): catalog discovery, declared primary keys, distribution metadata, row estimates from distribution statistics, bounded profiles and reads, labelled for cost attribution.
Available now Data warehouses
Shipped in the standard agent and backed by recorded conformance evidence.
Azure Synapse Analytics has no foreign keys. DataNivra cannot discover relationships from Azure Synapse Analytics: they come from an industry pack’s relationship mappings or relationships you declare. Without declared relationships, subsets do not follow relationships and the orphan check cannot cover them.
The console warns before a build that has none, and the certificate records RELATIONSHIPS_NOT_DECLARED for that source instead of reporting zero orphans as a check.
What works today
Dedicated and serverless SQL pool connector over TDS (the SQL Server connector's driver and TLS enforcement): catalog discovery, declared primary keys, distribution metadata, row estimates from distribution statistics, bounded profiles and reads, labelled for cost attribution.
SQL authentication with the password as a secret reference. TLS is always required and the certificate is verified; a session that is not encrypted is refused. Synapse is not SQL Server: it has its own connector entry and its own conformance record.
Authentication: SQL authentication (secret reference).
Known limitations
- Requires agent 0.4.1 or later: agent 0.4.0 cannot prove a dedicated-pool credential read-only and does not bind industry-pack relationship mappings to tables in your own schema (its driver, python-tds, ships in the standard agent image).
- Azure Synapse has no foreign keys, so DataNivra cannot discover relationships from Synapse: they come from an industry pack's relationship mappings or relationships you declare. Without declared relationships, subsets do not follow relationships and the orphan check cannot cover them (the certificate records RELATIONSHIPS_NOT_DECLARED).
- Pack relationship mappings bind to a table in your own schema (for example dbo.claims) only when exactly one table of that name exists in the dataset's sources and it has every column the mapping names; otherwise nothing binds and the console warns before the build. Relationships you declare yourself are approved identity-graph links between Synapse and another source; two Synapse tables are linked through a pack's relationship mappings.
- Primary keys are declared NOT ENFORCED: DataNivra reads them but cannot rely on Synapse to keep them unique.
- SQL authentication only; Microsoft Entra ID (service principal, managed identity) is not supported yet.
- Serverless pools keep no row statistics: counts and profiles read the underlying files and are billed per data processed.
- Run against a real serverless SQL pool (external tables over Parquet) and a real DW100c dedicated SQL pool. No emulator exists and SQL Server runs are not accepted as Synapse evidence.
- Serverless external tables declare no primary keys, so a dataset build rooted on one stops with ROOT_ENTITY_WITHOUT_PRIMARY_KEY (seen on the real run); the full workflow was proven on a dedicated pool.
- Serverless external tables also need REFERENCES on the database scoped credential they read through (see the least-privilege guide).
Capability matrix
Each capability is tracked separately and shown at the level the recorded conformance evidence supports. A connector that only discovers metadata is never presented as equivalent to one that runs the whole workflow.
| Capability | Level |
|---|---|
| Discovery | Supported |
| Classification | Supported |
| Profiling | Supported |
| Masking | Supported |
| Deterministic masking | Supported |
| Entity-aware subsetting | Supported |
| Relationship preservation | Supported |
| Synthetic workflows | Supported |
| Certification | Supported |
| Provisioning | Supported |
| CI/CD | Supported |
| Evidence generation | Supported |
Read-only proof
Proven from the pool's own permission functions before reading: the user must not be dbo or a member of db_owner, db_datawriter or db_ddladmin; fn_my_permissions must show no write, DDL, bulk-load or external-data permission on the database; HAS_PERMS_BY_NAME must show no INSERT, UPDATE, DELETE, ALTER or CONTROL on any visible table or view and no ALTER or CONTROL on any schema.
Conformance evidence
DataNivra runs every connector through a common conformance kit on synthetic data: no write statements, proven read-only access, secret redaction, schema and relationship discovery, bounded streaming, cancellation, clear reason codes, egress canaries and no raw values in logs, then the discover, mask, subset, certify and provision workflow. The published status is computed from those results.
| Engine or service | Version | Environment | Driver | Authentication | Verified on |
|---|---|---|---|---|---|
| Azure Synapse Analytics (dedicated SQL pool) | service | managed-service | python-tds 1.17.1 | password | 2026-10-12 |
| Azure Synapse Analytics (serverless SQL pool) | service | managed-service | python-tds 1.17.1 | password | 2026-10-12 |
Connector version 0.1.0, last verified 2026-10-12.
What crosses the boundary
Connectors run inside the agent in your network. Only metadata, aggregates and evidence reach DataNivra Cloud:
- Schema, table and column names and data types (for file sets, operator-declared or opaque table names — never data-derived file names).
- Aggregates such as row-count estimates, null and distinct ratios; counts from 1 to 10 are reported as 10 (small-cell protection).
- Classification findings (which column looks like which kind of sensitive data) and relationship metadata.
- Evidence metadata: checksums, gate outcomes, engine and policy versions, and templated error codes.
Never sent:
- Any cell value, row, sample, minimum/maximum or top-k value — masked or not.
- Credentials, connection strings or resolved secret values: DataNivra stores only the secret reference.
- Free text copied from your data, or exception messages that contain values.
See the customer-resident architecture and how to verify these claims yourself.
Setup outline
- Use agent 0.4.1 or later (agent 0.4.0 cannot prove a dedicated-pool user read-only or bind pack relationship mappings to your own schema).
- Create a SQL login and a database user with SELECT on the schemas to read and nothing else (not dbo, no db_owner, db_datawriter or db_ddladmin, no bulk-load or external-data permission). For serverless external tables, also grant REFERENCES on the database scoped credential they read through.
- Store the connection document with the password as a secret reference and set "pool" to "dedicated" or "serverless". TLS is always on and the certificate is verified.
- In the console, register the source with the secret reference of that document (for example vault://tdm/source-connection) and the agent that can reach it, then run discovery.
- Declare relationships: Azure Synapse has no foreign keys, so DataNivra cannot discover relationships from Synapse. Choose an industry pack whose relationship mappings match your tables (a mapping binds to a table in your own schema when exactly one table of that name has every mapped column) or approve identity-graph links. Without declared relationships, subsets do not follow relationships and the orphan check cannot cover them: the console warns before the build and the certificate records RELATIONSHIPS_NOT_DECLARED.
- Before reading, the agent proves the user is read-only from the pool’s own permission functions and refuses to continue if it is not. Serverless external tables declare no primary keys: a dataset build rooted on one stops with ROOT_ENTITY_WITHOUT_PRIMARY_KEY, so root subsets on a dedicated-pool table with a declared primary key.
Example connection document, stored in your secret store and resolved by the agent locally (DataNivra only ever sees its reference):
{
"kind": "AZURE_SYNAPSE",
"host": "example-ws.sql.azuresynapse.net",
"database": "claims_dw",
"pool": "dedicated",
"user": "tdm_reader",
"password_ref": "vault://tdm/claims-synapse#password",
"schemas": ["claims"]
}Next: install the agent, then follow the getting started guide.
All integrations · Data warehouse test data management · Guides · Start free