Exacs create dev db using backup

 

Intro

AI technologies are evolving at a rapid pace, but the landscape is still unstable. At the core of every AI system lies data, and validating AI models without proper test environments is simply not possible. To ensure reliability, lower environments must be refreshed regularly with production-scale data. Large datasets are essential for accuracy, and inadequate test data often leads to issues such as hallucinations something many enterprises are already struggling with.

Refreshing databases from higher environments is therefore a critical part of any testing strategy. However, managing large-scale refresh operations is complex and requires careful preparation to ensure scripts, processes, and dependencies are fully aligned. With Oracle ExaCS (Exadata Cloud@Customer), this process is greatly simplified, allowing DBAs to execute refresh operations efficiently and with minimal disruption—bringing environment refreshes quite literally to your fingertips.

Creating a Lower Environment Using EXACS

In this article, I’ll walk you through the process of creating a lower environment using Oracle Exadata Cloud@Customer (EXACS).

Scenario:

• Database Backups: Managed via the Automatic Recovery Service.

• Database Location: Databases are hosted on EXACS.

• Note: The Automatic Backup Service utilizes Zero Data Loss Recovery Appliance (ZDLRA) for restoring backups.

Step 1: Navigate to the Database Backup

Begin by accessing your database backup configurations to initiate the environment setup process. Once you locate the 

.

                                              Figure 1 : Navigate the backup

Step 2: Create the Database

This step is quite simple. Here, you’ll select the appropriate Pluggable Database (PDB) and the desired Availability Domain.

Make sure to carefully choose the correct Exadata Infrastructure, as selecting the wrong one can lead to deployment issues or connectivity problems later on.

                                               Figure 2 : Create database using backup

Step 3: Select the Correct Cluster

The next step is to select the appropriate cluster. Remember, ExaCS runs VMs on top of Exadata infrastructure, which allows you to create multiple VMs and segregate different environments. I’ll cover this topic in more detail in a separate blog.

                                        Figure 3 : Create database backup - 1  

Step 4 : Database Configuration and Duplication Setup

In Figure 4, you need to specify both the database name and the database unique name.

You can set the SYS password to any value that follows Oracle’s standard password conventions.

For a database duplication, ensure that you provide the TDE password, as the backup pieces will be directly retrieved from ZDLRA.

                                                  Figure 4 : Set passwords.  

An interesting piece of information I’d like to share is where to find the duplicate database logs.

1. Log in to the target host where the duplicate database is running.

2. The duplicate logs are located at: /var/opt/oracle/log/TEST2/dtrs/rman/bkup

3. Sample log file: DBRS_IRESTORE_2025-08-08_18-41-56-5774831298590343329.log

[grid@exatestdb-node1 bkup]$ tail -200f DBRS_IRESTORE_2025-08-08_18-41-56-5774831298590343329.log
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467743375_198"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467743401_198"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467734225_219"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467734315_219"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467734317_219"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467734307_219"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467734316_219"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467734322_219"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467743306_254"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467743388_254"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467729735_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467730363_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467730614_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467730825_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467730875_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467730901_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467730919_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467730923_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467730940_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467730986_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467730985_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467731010_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467731023_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467731036_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467731836_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467731341_271"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467743414_273"
new media label is "Recovery Appliance (raiadp4)" for piece "VB$_3990092206_467743537_273"
channel RA0: starting datafile backup set restore
channel RA0: specifying datafile(s) to restore from backup set
channel RA0: restoring datafile 00009 to +DATAC1/TEST_EXA/DATAFILE/reefer_svc_tbs.2509.1208630651
channel RA0: restoring section 1 of 18
channel RA0: reading from backup piece VB$_3890646435_481469587_9
channel RA1: starting datafile backup set restore
channel RA1: specifying datafile(s) to restore from backup set
channel RA1: restoring datafile 00010 to +DATAC1/TEST_EXA/DATAFILE/generic_sensor_hist_2024_q3.3874.1208630651
channel RA1: restoring section 1 of 10
channel RA1: reading from backup piece VB$_3890646435_481469600_10
channel RA2: starting datafile backup set restore
channel RA2: specifying datafile(s) to restore from backup set
channel RA2: restoring datafile 00011 to +DATAC1/TEST_EXA/DATAFILE/generic_sensor_hist_2024_q1.2488.1208630651
channel RA2: restoring section 1 of 10
channel RA2: reading from backup piece VB$_3890646435_481469605_11
channel RA3: starting datafile backup set restore

Conclusion

AI systems rely heavily on data, making well-maintained test environments essential for validating models and avoiding issues like hallucinations. Refreshing lower environments with production-scale data is critical but complex, requiring careful coordination of scripts, dependencies, and processes. Oracle Exadata Cloud@Customer (ExaCS) simplifies this, enabling DBAs to perform efficient, secure, and low-downtime environment refreshes.