How to Migrate a Windows Oracle RAC to Linux Using DataGuard: Step‑by‑Step Guide

Background

The article records a migration from a Windows‑based Oracle RAC (two‑node, 11.2.0.4, about 2.5 TB of data) to a Linux RAC, with a target downtime of roughly 15 minutes. Only the steps for building DataGuard from a single‑instance database to a RAC are covered.

Implementation Steps

1. Preparation

Perform preliminary database configuration checks, such as confirming that archive logging is enabled.

Single‑instance primary database IP: 192.168.100.100

RAC node IPs: 192.168.100.101/102, VIPs: 103/104, SCAN IP: 105

2. Parameter Files

The primary database uses online modification of its parameter file. Important note: log_file_name_convert does not map the path under db_recovery_file_dest because the online redo logs are automatically mapped to the standby location.

If data files are scattered, map all paths to +data/orcl/datafile for easier management.

After editing, rename the file to initorcl1.ora and place it in $ORACLE_HOME/dbs.

3. Modify tnsnames.ora

Update the primary single‑instance tnsnames.ora with the RAC connection details and copy the file to all standby nodes.

4. Password File

Copy the primary password file orapw<$ORACLE_SID> to each standby node and rename it to orapworcl1 and orapworcl2.

5. Full Backup on Primary

Take a full RMAN backup of the primary database and copy it to standby node 1. Ensure the primary backup schedule is stopped and the RMAN ARCHIVELOG DELETION POLICY is set to applied on standby.

6. Backup Standby Control File

After the data files are transferred, back up the standby control file on the primary.

7‑8. Transfer and Start Standby Instance

Copy the backup control file to standby node 1, then start the standby instance in NOMOUNT state using the standby parameter file.

9. Restore Standby Control File

Use RMAN on standby node 1 to restore the standby control file.

10‑11. Register Backup Set and Restore Data Files

Register the backup set on the standby and restore the data files, then start the listener on standby node 1 (using netca or netmgr). The instance registers automatically with the default service name orcl_st.

12‑13. Start MRP and Monitor

Start the Managed Recovery Process (MRP) on standby node 1 (mount state) and monitor the alert.log to track synchronization progress.

14. Final Steps

After incremental recovery finishes, stop MRP, open the standby database, and restart MRP.

The complete procedure sets up DataGuard from a single‑instance database to an Oracle RAC, with additional post‑migration tasks such as adding nodes, configuring backup and archive‑log deletion policies, which are not detailed here.