Integrations · Object storage · Coming next
Google Cloud Storage: coming next connector
File-set connector for Parquet, CSV or JSON-lines objects under a Cloud Storage bucket prefix, read through the Cloud Storage JSON API with bounded listings and ranged reads, into the same discovery, masking, subsetting and certification path as the other object stores.
Coming next Object storage
Active implementation; not yet available for production use.
Not available yet. Active implementation; not yet available for production use. The console does not offer it, and asking an agent for a connector it does not ship fails with the error code CONNECTOR_NOT_AVAILABLE; the product never fakes a connection.
What you can do today
Copy extracts to local files or an S3-compatible store read by a shipped connector, or use Cloud Storage's S3-compatible XML API with HMAC keys through the Amazon S3 / S3-compatible connector (read-only by attestation).
Whatever route you choose, the export and every row-level step stay inside your network. You can evaluate the whole workflow now with the synthetic demo.
Known limitations
- Requires agent 0.4.3 or later: earlier agents do not include the connector (its only driver dependency, google-auth, is already in the standard agent image).
- The read-only proof covers IAM on the bucket (including legacy bucket roles); per-object ACLs of fine-grained buckets are not visible to it, so use uniform bucket-level access.
- A listing larger than max_list_objects (default 100,000 objects) and objects larger than max_object_bytes (default 10 GiB) are refused rather than read partially.
- An access token is not refreshed: it must outlive the job. Service-account keys and workload identity refresh automatically.
- Requester-pays buckets and customer-supplied encryption keys are not supported.
Capability matrix
Google Cloud Storage does not ship yet, so no capability is verified. The column shows what the design plans; plans are not commitments.
| Capability | Design intent |
|---|---|
| Discovery | Planned |
| Classification | Planned |
| Profiling | Planned |
| Masking | Planned |
| Deterministic masking | Planned |
| Entity-aware subsetting | Planned |
| Relationship preservation | Planned |
| Synthetic workflows | Planned |
| Certification | Planned |
| Provisioning | Planned |
| CI/CD | Planned |
| Evidence generation | Planned |
How the connector is designed to prove read-only access
Proven before reading with the bucket's testIamPermissions: the credential must hold storage.objects.get and storage.objects.list and none of storage.objects.create, delete, update or setIamPolicy, storage.buckets.update, delete or setIamPolicy, multipart uploads, managed folder or hierarchical-namespace folder changes. If the permission check cannot be answered the source is refused; there is no attestation path.
Grant a dedicated service account only the Storage Object Viewer role on the bucket. Tokens are requested with the devstorage.read_only scope, the HTTP client can send only GET and HEAD to the Cloud Storage endpoint, and TLS is verified by default.
Planned authentication: service-account key (secret reference); workload identity / application default credentials; OAuth access token (secret reference).
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.
Request this connector
We prioritise connectors by demand, engineering effort and strategic fit. Telling us that Google Cloud Storage matters to you is counted; it does not commit a delivery date.
Need something specific from the Google Cloud Storage connector? Request a feature (please describe it without real data).
All integrations · Object storage test data management · Guides · Start free