Migrate an on-premises Oracle database to Amazon RDS for PostgreSQL by using an Oracle bystander and AWS DMS

Set up AWS SCT.

Create a new report, and connect to Oracle as the source and PostgreSQL as the target. In Project Settings, go to the SQL Scripting tab. Change the Target SQL Script to Multiple Files. These files will be used later and named the following:

Convert the Oracle database schema.

In the Action tab, choose Generate Report. Then, choose Convert Schema and choose Save as SQL.

Modify the scripts.

For example, you might want to modify the script if a number in the source schema has been converted to numeric format in PostgreSQL, but you want to use BIGINT instead for better performance.

Create the Amazon RDS DB instance.

In the correct AWS Region, create a new PostgreSQL DB instance. For more information, see Creating a PostgreSQL DB instance and connecting to a database on a PostgreSQL DB instance in the Amazon RDS documentation.

Configure DB instance specifications.

Specify the DB engine version, DB instance class, Multi-AZ deployment, storage type, and allocated storage. Enter the DB instance identifier, a primary user name, and a primary password.

Configure network and security.

Specify the virtual private cloud (VPC), subnet group, public accessibility, Availability Zone preference, and security groups.

Configure database options.

Specify the database name, port, parameter group, encryption, and KMS key.

Configure backups.

Specify the backup retention period, backup window, start time, duration, and whether to copy tags to snapshots.

Configure monitoring options.

Activate or deactivate enhanced monitoring and performance insights.

Configure maintenance options.

Specify auto minor version upgrade, maintenance window, and start day, time, and duration.

Run the pre-migration scripts from AWS SCT.

On the Amazon RDS instance, run the following scripts generated by AWS SCT:

Set up the network for Amazon EC2.

Create the new VPC, subnets, internet gateway, route tables, and security groups.

Create the EC2 instance.

In the appropriate AWS Region, create a new EC2 instance. Select the Amazon Machine Image (AMI), choose the instance size, and configure instance details: number of instances (1), the VPC and subnet you created in the previous task, auto-assign public IP, and other options. Add storage, configure security groups, and launch. When prompted, create and save a key pair for the next step.

Connect the Oracle source database to the EC2 instance.

Copy the IPv4 public IP address and DNS to a text file and connect by using SSH as follows: ssh -i "your_file.pem" ec2-user@<your-IP-address-or-public-DNS>.

Set up the initial host for a bystander in Amazon EC2.

Set up SSH keys, bash profile, ORATAB, and symbolic links. Create Oracle directories.

Set up the database copy for a bystander in Amazon EC2

Use RMAN to create a database copy, enable supplemental logging, and create the standby control file. After copying is complete, place the database in recovery mode.

Set up Oracle Data Guard.

Modify your listener.ora file and start the listener. Set up a new archive destination. Place the bystander in recovery mode, replace temporary files to avoid future corruption, install a crontab if necessary to prevent the archive directory from running out of space, and edit the manage-trclog-files-oracle.cfg file for the source and standby.

Prep the Oracle database to sync shipping.

Add the standby log files and change the recovery mode. Change the log shipping to SYNC AFFIRM on both the source primary and the source standby. Switch logs on primary, confirm via the Amazon EC2 bystander alert log that you are using the standby log files, and confirm that the redo stream is flowing in SYNC.

Create a replication instance in AWS DMS.

Complete the fields for the name, instance class, VPC (same as the Amazon EC2 instance), Multi-AZ, and public accessibility. Under Advance, specify allocated storage, subnet group, Availability Zone, VPC security groups, and AWS Key Management Service (AWS KMS) key.

Create the source database endpoint.

Specify the endpoint name, type, source engine (Oracle), server name (Amazon EC2 private DNS name), port, SSL mode, user name, password, SID, VPC (specify the VPC that has the replication instance), and replication instance. To test the connection, choose Run Test, and then create the endpoint. You can also configure the following advanced settings: maxFileSize and numberDataTypeScale.

Connect AWS DMS to Amazon RDS for PostgreSQL.

Create a migration security group for connections across VPCs.

Create the target database endpoint.

Specify the endpoint name, type, source engine (PostgreSQL), server name (Amazon RDS endpoint), port, SSL mode, user name, password, database name, VPC (specify the VPC that has the replication instance), and replication instance. To test the connection, choose Run Test, and then create the endpoint. You can also configure the following advanced settings: maxFileSize and numberDataTypeScale.

Create the AWS DMS replication task.

Specify the task name, replication instance, source and target endpoints, and replication instance. For migration type, choose Migrate existing data and replicate ongoing changes. Clear the Start task on create checkbox.

Configure the AWS DMS replication task settings.

For target table preparation mode, choose Do nothing. Stop task after full load completes (to create primary keys). Specify limited or full LOB mode, and activate control tables. Optionally, you can configure the CommitRate advance setting.

Configure table mappings.

In the Table mappings section, create an Include rule for all tables in all schemas included in the migration, and then create an Exclude rule. Add three transformation rules to convert the schema, table, and column names to lowercase, and add any other rules needed for this specific migration.

Start the task.

Start the replication task. Make sure that the full load is running. Run ALTER SYSTEM SWITCH LOGFILE on the primary Oracle database to kick-start the task.

Run the mid-migration scripts from AWS SCT.

In Amazon RDS for PostgreSQL, run the following scripts generated by AWS SCT:

Restart the task to continue change data capture (CDC).

Run VACUUM on the Amazon RDS for PostgreSQL DB instance, and restart the AWS DMS task to apply cached CDC changes.

Review the AWS DMS logs and validation tables for any errors.

Check and fix any replication or validation errors.

Stop all Oracle dependencies.

Stop all Oracle dependencies, shut down listeners on the Oracle database, and run ALTER SYSTEM SWITCH LOGFILE. Stop the AWS DMS task when it shows no activity.

Run the post-migration scripts from AWS SCT.

In Amazon RDS for PostgreSQL, run the following scripts generated by AWS SCT:

Complete additional Amazon RDS for PostgreSQL steps.

Increment sequences to match Oracle if needed, run VACUUM and ANALYZE, and take a snapshot for compliance.

Open the connections to Amazon RDS for PostgreSQL.

Remove the AWS DMS security groups from Amazon RDS for PostgreSQL, add production security groups, and point your applications to the new database.

Clean up AWS DMS objects.

Remove the endpoints, replication tasks, replication instances, and the EC2 instance.