Supported Data Warehouses
AI-Link enables users to connect business-curated data from AtScale Semantic Models to downstream analytic and AI workloads. As such, use of AI-Link is dependent upon the AtScale Semantic Layer being installed and configured. AI-Link 1.0.0->2.x.x is designed to work with the installer-based version of AtScale.
The tables below outline the functionality that AI-Link supports for each warehouse. Read functionality fetches data via the DataModel object, while write functionality uses the DataModel object to write data back to the warehouse.
General Capabilities
| Managed Features | EDA/Python Helper Functions | Semantic Predictions | Semantic Inference (UDF) | Auto Semantic Model Creation |
|---|
| Snowflake | ☑ | ☑ | ☑ | ☑ | ☑ |
| Databricks | ☑ | ☑ | ☑ | ☑ | ☑ |
| GBQ | ☑ | ☑ | ☑ | ☑ | ☑ |
| Iris | ☑ | ☑ | ☑ | | |
| Synapse | ☑ | ☑ | ☑ | | |
| Redshift | ☑ | ☑ | | | |
| Spark | ☑ | ☑ | ☑ | | |
| Hadoop | ☑ | ☑ | | | |
| PostgreSQL | ☑ | ☑ | ☑ | | |
Read Functionality
| get_data | get_data_direct | get_data_jdbc | get_data_spark_jdbc | get_data_spark |
|---|
| Snowflake | ☑ | ☑ | ☑ | ☑ | |
| Databricks | ☑ | ☑ | ☑ | ☑ | ☑ |
| GBQ | ☑ | ☑ | ☑ | | |
| Iris | ☑ | ☑ | ☑ | | |
| Synapse | ☑ | | ☑ | | |
| Redshift | ☑ | | ☑ | ☑ | |
| Spark | ☑ | ☑ | ☑ | ☑ | |
| Hadoop | ☑ | ☑ | ☑ | | |
| PostgreSQL | ☑ | ☑ | ☑ | | |
Write Functionality
| join_udf | write_feature_importance | writeback | writeback_spark_jdbc | writeback_spark | write_df_to_db |
|---|
| Snowflake | ☑ | ☑ | ☑ | ☑ | | ☑ |
| Databricks | ☑ | ☑ | ☑ | ☑ | ☑ | ☑ |
| GBQ | ☑ | ☑ | ☑ | ☑ | | ☑ |
| Iris | | ☑ | ☑ | ☑ | | ☑ |
| Synapse | | ☑ | ☑ | ☑ | | ☑ |
| Redshift | | | | | | |
| Spark | | ☑ | ☑ | ☑ | ☑ | ☑ |
| Hadoop | | | | | | |
| PostgreSQL | | ☑ | ☑ | | | ☑ |