Structured data stores¶
Structured data is highly organized and follows a strict schema or "blueprint". Examples of this include numeric, dates, strings, and boolean values.
Connector vs. Driver
Each data store connects to DataRobot through either a driver or a connector, and is indicated in the tables for supported unstructured and structured data stores.
| Type | Description | Recommended use |
|---|---|---|
| Connector | DataRobot connects through the data store's native API using an integration built and maintained by DataRobot. The connection is defined by connection fields instead of a JDBC URL, supports authentication methods such as OAuth and service principals, and, for many data stores, supports unstructured data. Connectors require a registered data store. | Recommended when you want to configure connections and credentials once and reuse them. They are also the preferred option when credentials or other sensitive configuration are stored in DataRobot and should not be embedded in your code. Connectors are also recommended for all agentic workflows due to lower latency and unstructured data support. |
| Legacy JDBC Driver | DataRobot connects using the JDBC driver provided by the database vendor. The connection is defined by a JDBC URL and JDBC parameters, authenticates with basic credentials, and supports structured (tabular) data. For Self-Managed AI Platform installations, an administrator uploads and manages the driver files. Requires a registered data store. | Recommended only for custom connection types that are not supported by connectors or a DataRobot-managed JDBC driver. |
| DataRobot-managed JDBC driver | DataRobot ships and manages the driver set as part of the install (bundled in the browser-jdbc service). There are three main differences from the legacy driver:
This driver currently supports Postgres, MySQL, SQL Server, BigQuery, Snowflake, Databricks, Trino, SAP Datasphere drivers. |
Recommended when you are prepared to manage the connection configuration and credentials directly in the your code. |
The following table lists all data stores—both native connectors and JDBC drivers—supported by DataRobot for structured data ingest:
* Only supported with JDBC, not with KDB native query language q
Self-Managed AI Platform
See Manage JDBC drivers for steps to upload JDBC drivers to your organization.
Deprecated connections¶
Support is deprecated for these drivers and connectors:
| Connection | Connection type | Version |
|---|---|---|
| ADLS Gen2 | JDBC | 2021.2.1634676262008; 2020.3.1605726437949 |
| Apache Hive * | JDBC | All |
| Amazon S3 | JDBC | 2020.3.1603724051432 |
| Elasticsearch | JDBC | All |
| Google BigQuery | JDBC | spark-1.2.23.1027 |
| Microsoft SQL Server | JDBC | 6.0 |
* The Apache Hive JDBC driver is only deprecated for multi-tenant SaaS installations.
Note
Older driver versions may still exist, but DataRobot recommends that you use the latest supported version of a connection.