Skip to content

Delta Lake

Connect a Delta Lake table for federated analysis. The table is read at its storage location and can be joined with datasets from other connectors.

How Delta is addressed

A Delta connection points at one Delta table by its path. Unlike Iceberg, there is no separate catalog service — the table path is the address, and the transaction log at that path defines the current table state.

The path can be:

  • An object-storage location, e.g. s3://bucket/delta-table/. Provide the storage credentials to read it.
  • A local filesystem path — available in the desktop app and self-hosted deployments, not on the managed service (see Managed vs. self-hosted).

Before you begin

Connection settings

Choose the storage location first, then provide the path:

FieldRequiredDescription
Storage locationYesObject storage, or a local path where the environment permits it.
Delta table pathYesThe table's root, e.g. s3://bucket/delta-table/.

For an object-storage path, also provide the object-storage access credentials — vendor preset, endpoint, region, keys, URL style, SSL.

Test and save

Enter the fields and use Test connection. On success, the table is available for analysis.

Capabilities

  • Execution model: Federated. The Delta table can be joined with datasets from other connectors.
  • Access: Read-only. The current snapshot defined by the transaction log is read.
  • Backends: S3-compatible object storage; local paths on desktop / self-hosted only.

Troubleshooting

  • Table not found — verify the path points at the table root (the directory containing the transaction log), not a single data file.
  • Access denied — for an object-storage path, confirm the credentials grant read access and the region is correct.
  • Credential fields are hidden — they appear only when the storage location is set to object storage; a local path does not use them.

Datyo · The Agent-native intelligent data analytics platform