Internal Sources

Oracle AI Data Platform supports ingestion from internal Oracle sources using built-in ingestion connectors. These connectors enable users to seamlessly extract data using Spark-based notebooks and integrate it into their workflows and data pipelines.

Ingestion connectors abstract the complexities of connection setup, providing optimized access patterns for both batch and near real-time ingestion from Oracle-native services.

AI Data Platform provides sample code templates in this ZIP file to support ingesting data from several internal sources using Spark in notebooks.

Table 14-1 Internal Sources

Source Access Type Integration Method Decription External Catalog Support Sample Code Available
Fusion Extract Only Preconfigured Spark Templates Extracts data from Fusion SaaS applications via BICC into AI Data Platform tables or volumes. No Yes
REST Endpoints Read Only JDBC via Spark Notebook Reads from APIs for ingesting semi-structured data like JSON. No Yes
MySQL HeatWave Read Only JDBC via Spark Notebook Move data between AI Data Platform and MySQL HeatWave using JDBC. No Yes
Autonomous Data Warehouse (ADW) Read/Write + Zero-Copy JDBC or External Catalog Ingest from or register ADW as an external catalog for querying data directly without duplication. Yes Yes
Autonomous Transaction Processing Read/Write + Zero-Copy JDBC or External Catalog Ingest from or register ADW as an external catalog for querying data directly without duplication. Yes Yes
Oracle DB Read/Write JDBC or External Catalog Supports data ingestion from on-prem or OCI Oracle Databases. Yes Yes
Exadata Read/Write JDBC or External Catalog Access Exadata systems for high-performance reads and writes using JDBC. No Yes

Table 14-2 Spark SQL to Oracle Database, Autonomous Database, Exadata Data Type mapping

Spark SQL Type Oracle Database, Autonomous Database, Exadata Data Type
ByteType NUMBER(38,10)
ShortType NUMBER(38,10)
IntegerType (INT) NUMBER(38,10)
LongType NUMBER(38,10)
FloatType FLOAT(126)
DoubleType NUMBER(38,10)
DecimalType(p,s) NUMBER(p,s)
StringType VARCHAR2(4000 CHAR)
BinaryType BLOB
BooleanType VARCHAR2(4000 CHAR)
DateType DATE
TimestampType TIMESTAMP(6)
ArrayType VARCHAR2(4000 CHAR)
MapType Not supported
StructType VARCHAR2(4000 CHAR)
CalendarIntervalType Supported if converted to String/VARCHAR2