Oracle Autonomous AI Database Considerations

This topic highlights some important aspects of using Oracle Autonomous AI Database with Resource Analytics.

Usage Guidelines for Oracle Autonomous AI Database Associated with Resource Analytics

Resource Analytics provisions instances of Oracle Autonomous AI Database (ADW) to store data. As part of Resource Analytics, Oracle doesn't let you perform, or recommend, certain Autonomous AI Database administration tasks.

As a service administrator, you can access the Oracle Autonomous AI Database instance associated with Resource Analytics using the Oracle Cloud Infrastructure Console. The following table outlines the tasks that aren't allowed or recommended.


Task	More information
Database lifecycle management	Although an administrator can stop, pause, and restart the Autonomous AI Database (ADW) associated with a Resource Analytics instance, Resource Analytics doesn't update any resource metadata while that ADW is stopped, paused, or restarting. If it's stopped or paused for more than 21 days, Resource Analytics permanently stops updating it. We recommend that you not rename the ADW. Note: Never delete the ADW without first deleting the Resource Analytics instance that's tied to it.
Changes to auto-provisioned resources	You can scale up resources such as OCPUs for more capacity or terabytes for extra storage. However, you can't scale down such resources to levels under those automatically provisioned with Resource Analytics. Autoscale for OCPUs is turned on by default by Resource Analytics and you can't turn it off.
Autonomous Data Guard	Even though you can turn on Autonomous Data Guard, this isn't recommended because Resource Analytics doesn't support failover to the instance populated through Data Guard.
Restore from backups	Autonomous AI Database provides daily backups. While restoring to a new instance might be useful to view or copy data from backup, it isn't recommended to restore backups against the Autonomous AI Database instance associated with Resource Analytics. Doing so might put the data pipelines and factory content in an inconsistent state.

The Autonomous AI Database instance created by Resource Analytics is shared between you and the service. Be aware of the impact certain user actions and configurations can have on the performance of queries and service-provided updates. While you can insert custom data and query the database, it has an impact on the database storage and compute. The following are some guidelines:


Custom Usage	Guidance
Adding data to custom schemas	You can add as many as 50 GB of custom data to the service. For more than 50 gigabytes, you must scale up the storage through the Oracle Cloud Infrastructure Console, which is charged to your Oracle Cloud account. If you intend to do compute-intensive queries and transformations on your custom data, then you can scale up more OCPUs using the Oracle Cloud Infrastructure Console, which is charged to your Oracle Cloud account. See Scale Up Oracle Autonomous Data Warehouse.
Adding extra CPUs	Ensure that you have adequate CPUs configured in Autonomous AI Database (and corresponding sessions) to run custom ETL. For every CPU that's allocated as part of the Resource Analytics instance creation, you shouldn't exceed five low service concurrent sessions for custom ETL. Exceeding this can impact the ETL or reporting performance and therefore isn't supported. If you need to run more sessions, then ensure that extra CPUs are added to the Autonomous AI Database instance. Each extra CPU that you add this way gives as many as 300 low sessions. You must avoid Medium or High sessions in all cases because they block the ETL pipeline.
Adding database connections	See Tutorial. Follow these guidelines: Aim for fewer than 10 concurrent database connections. When you configure the connections, ensure that you select the service with the low suffix in its name. Don't select the service with the high prefix because it can lead to performance issues in reports and delays in completion of daily data refreshes in Resource Analytics. Close unused sessions in a timely manner to ensure no adverse impact on the performance of Resource Analytics. If you plan to run compute-intensive queries or create many concurrent database connections, then you can scale up extra OCPUs through the Oracle Cloud Infrastructure Console, which is charged to your Oracle Cloud account.

Warning

Resource Analytics provisions your Autonomous AI Database instance with Private Endpoint access. This Private Endpoint access must remain in place to apply resource updates.

If you change Database Network Settings, Private Endpoint access must remain enabled. Do not switch to another access type. Modifying or removing the Private Endpoint will disrupt resource metadata transfers and database updates. See: Enable Public Access with Private Endpoints.

Provisioning the Required Autonomous AI Database Instance

The Autonomous AI Database instance is provisioned when you create the Resource Analytics instance.

You need to provide some basic information to let Resource Analytics provision the required ADW instance:

Either an administrator password or details of an appropriate secret contained within your OCI vault
The network to be used (VNC and subnet)

If a text-based password is supplied, the username can't be changed, but you're free to enter any valid ADW password. (The specification for this password is 12 to 30 characters, containing at least one uppercase letter, one lowercase letter, and one numeric value. Spaces aren't allowed.)

Note

Any password provided isn't stored by Resource Analytics.

You can change the password in the ADW service Console when ADW has completed provisioning.

For more information, see Creating an Instance.

Starting and Stopping the ADW: Implications for Resource Analytics

Autonomous AI Database is a critical component of the Resource Analytics instance. Stopping or disabling the database has direct implications on the functionality and data freshness of your Resource Analytics setup.

What Happens When the Autonomous AI Database is Stopped?

Live updates are paused
The Resource Analytics instance stops receiving real-time updates related to resource changes.
The instance state changes to Needs Attention
This indicates that user action is needed.

How to Recover

Re-enable or re-start the ADW instance to resolve the issue. When it's active, the Resource Analytics pipeline resumes processing updates within an hour.

Note

Consider the following:

Backfill delay: On ADW restart, the processing and ingesting any accumulated (pending) data events - a process known as backfill - might take extra time to complete. As a result, you might experience delays in data refresh and availability within the system.

Impact of extended downtime: If the ADW instance remains shut down for an extended period, the duration required to backfill pending events increases proportionally.

Automatic ingestion suspension

If the ADW instance is offline for more than 21 consecutive days, automatic data ingestion is permanently disabled for the associated Resource Analytics instance. At this point:

The affected Resource Analytics is marked as obsolete.
The only solution available is to delete the obsolete Resource Analytics instance and create a new one.

Plan the maintenance activities to minimize data pipeline disruption and ensure continued service availability.