Prerequisites for Creating a Connection

Satisfy the following prerequisites appropriate to your environment for creating a connection with Oracle Integration.

Download the Credentials Wallet and Obtain Database Details

You must download the credentials wallet and obtain database details.

  1. Download the client credentials wallet from the Oracle Autonomous AI Transaction Processing instance. See Download Client Credentials (Wallets) in Using Oracle Autonomous AI Database Serverless.
  2. Ensure that the target database is accessible.
  3. Ensure that you have write permissions to the database.
  4. Ensure that you have the required permissions to run stored procedures and SQL statements.
  5. Know the database service name.
  6. Know the database service user name and password for connecting to the database.

Satisfy Bulk Data Import Operation Prerequisites

If you want to use the bulk data import operation, you must configure the Oracle Autonomous AI Transaction Processing (ATP) Adapter to use the JDBC With OCI Signature security policy on the Connections page. Before you can successfully configure this security policy on the Connections page, you must complete the following prerequisites.

  1. Create an object storage bucket in Oracle Cloud Infrastructure and obtain the OCID value of the compartment in which you create the storage bucket. You need this value when configuring the security policy on the Connections page.

    You specify the OCID of the object storage user on the Connections page when configuring the JDBC With OCI Signature security policy. This user should have the manage bucket and manage object privileges. Set the following policies to manage privileges on the bucket and object in the object storage:

    Allow group group_name to manage buckets in compartment compartment_name
    
    Allow group group_name to manage objects in compartment compartment_name 
    

    See How Policies Work and Policy Reference.

  2. Obtain the tenancy OCID and user OCID details in the Oracle Cloud Console. You need these values when configuring the security policy on the Connections page. When you sign up for Oracle Cloud Infrastructure, Oracle creates a tenancy for your company, which is a secure and isolated partition within Oracle Cloud Infrastructure where you can create, organize, and administer your cloud resources.
    1. Sign in to the Oracle Cloud Console.
    2. Open the navigation menu and click Governance & Administration. Under Account Management, click Tenancy Details.
    3. In the Tenancy information section, click Show to display the OCID tenancy value.
    4. Copy the value. You need this value when configuring the security policy on the Connections page.
    5. Click the Profile icon and select User Settings.

      Note:

      You can also open the navigation menu and click Identity & Security. Under Identity, click Users to access the user profile.
    6. Click Show to display the OCID user value.
    7. Copy the value.
  3. Create an API signing key. You then specify the signing key in Oracle Cloud Infrastructure.
    1. Sign in to the Oracle Cloud Console.
    2. Open the navigation menu and click Identity & Security. Under Identity, click Users.
    3. On the Users page, click the link of the user name to use.
    4. Under Resources, click API Keys, then click Add API Key.

      Note:

      Only a private key without a pass phrase/password is supported.
    5. In the Add API Key dialog, select Paste Public Key and enter the contents of the public key you created, then click Add.
    6. Copy the finger print value generated by Oracle Cloud Infrastructure. You need this value when configuring the JDBC With OCI Signature security policy on the Connections page.

      API Keys section with an Add Public Key link and fingerprint value.

Manage AI and Vector Profiles

You must create and manage your AI profiles and vector profiles.

  • AI profiles: Create and manage your AI profiles through the DBMS_CLOUD_AI package.

    See Manage AI Profiles.

  • Vector profiles: Create and manage your vector profiles and index through the DBMS_CLOUD_AI package.

    See CREATE_VECTOR_INDEX Procedure.

Migrate from an Oracle Database Classic Cloud Service Database Instance to an Oracle Autonomous AI Transaction Processing or Oracle Autonomous Data Warehouse Database Instance

Perform the following steps if you want to migrate from an Oracle Database Classic Cloud Service database instance to an Oracle Autonomous AI Transaction Processing or Oracle Autonomous Data Warehouse database instance.

  1. Migrate all the required database objects, stored procedures, wrapper procedures, and tables to the destination Oracle Autonomous AI Transaction Processing or Oracle Autonomous Data Warehouse database instance.
  2. Change the Oracle Database Cloud Service Adapter connection details to point to an Oracle Autonomous AI Transaction Processing or Oracle Autonomous Data Warehouse database instance.
    1. Go to the Connection page for the Oracle Database Cloud Service Adapter.
    2. Click Configure Connectivity.
    3. Specify the new host name.
    4. Specify the new service name and click OK.
    5. Click Configure Security.
    6. Select the Oracle Wallet security policy.
    7. Upload the wallet.
    8. Specify the wallet password and reconfirm it.
    9. Specify the database service username.
    10. Specify the database service password, reconfirm it, and click OK.
    11. Delete the agent from the connection.
  3. Test the connection.
  4. Once the test is successful, click Save to save the connection details.
  5. Reactivate the integrations.

Configure the Embedding Models and Database User Privileges

You must configure the embedding models and database user privileges.

Configure the Embedding Models in Oracle Autonomous AI Transaction Processing

Vector ingestion and vector search in Oracle Autonomous AI Transaction Processing require pretrained embedding models to be preconfigured and available at the database level. These models generate vector embeddings from document content and are mandatory for all vector-based operations.

Before performing vector ingestion, upload one or more approved embedding models to the database. Embedding models configured as part of this prerequisite are reused across document ingestion pipelines, vector search queries, and hybrid search queries.

Note:

Vector ingestion or search cannot be executed unless this prerequisite is satisfied.

See Convert Pretrained Models to ONNX Model: End-to-End Instructions for Text Embedding.

Use DBMS_DATA_MINING.IMPORT_ONNX_MODEL to import the embedding model.

Sample Query: DBMS_DATA_MINING.IMPORT_ONNX_MODEL
DBMS_DATA_MINING.IMPORT_ONNX_MODEL(
  'EMBEDDING_MODEL_NAME',
  EMBEDDING_MODEL_SOURCE_BLOB,
  JSON(
    '{
      "function": "embedding",
      "input": {
        "input": ["INPUT_ATTRIBUTE"]
      },
      "embeddingOutput": "OUTPUT_ATTRIBUTE"
    }'
  )
);

Grant Database User Privileges

Grant the database user used by the integration the privileges required for hybrid index operations. These privileges support table creation, vectorizer preference creation, hybrid vector index creation, vector search package execution, and hybrid index ingestion, search, and ingestion status operations.

Run the following SQL statements as a DBA or administrator user. Replace db_user with the Oracle Autonomous AI Transaction Processing database user used by the adapter connection.
GRANT CONNECT TO db_user;
GRANT RESOURCE TO db_user;
GRANT CREATE SEQUENCE TO db_user;
GRANT CREATE TRIGGER TO db_user;
GRANT CREATE TABLE TO db_user;
GRANT CREATE JOB TO db_user;
GRANT EXECUTE ON DBMS_VECTOR_CHAIN TO db_user;
GRANT EXECUTE ON DBMS_SCHEDULER TO db_user;
GRANT EXECUTE ON CTX_DDL TO db_user;
GRANT EXECUTE ON DBMS_CLOUD TO db_user;
GRANT EXECUTE ON DBMS_HYBRID_VECTOR TO db_user;
ALTER USER db_user QUOTA UNLIMITED ON DATA;