About Dedicated Elastic Pools

Dedicated elastic pools let you group Autonomous AI Database instances in an elastic pool where the pool leader and all pool members are physically co-located on the same dedicated infrastructure. This model is well suited to SaaS applications, microservices, and other environments that run a large number of relatively smaller databases while requiring predictable shared capacity and centralized operational control.

In addition to the standard elastic pool benefits, dedicated elastic pools provide a simplified pool-level storage billing and greater control over Database, Oracle REST Data Services (ORDS), and Oracle APEX maintenance. The pool leader manages pool capacity, including compute and storage scaling, and controls maintenance scheduling and patching for the pool. Each member retains its own database allocation while drawing from the pool’s available compute and storage capacity.

Before You Create or Join a Dedicated Elastic Pool

Requirements for a Pool Leader

To create a dedicated elastic pool and become a pool leader, an Autonomous AI Database instance must:

Use the ECPU compute model.
Have compute and storage auto scaling disabled.
Not be a member of an existing elastic pool.
Have sufficient number of ECPUs and storage available, within the tenancy limit, to accommodate the elastic pool's compute and storage size.

Provisioning a pool leader is subject to tenancy-level service limits.

The maximum allowed individual ECPU count for an Autonomous AI Database instance that creates an elastic pool is four (4) times the pool size specified while creating the pool.

Requirements of a Pool Member

To join a dedicated elastic pool, an Autonomous AI Database instance must:

Use the ECPU compute model.
Have compute and storage auto scaling disabled.
Not be a member of an elastic pool.
Have an ECPU count below the available pool capacity.

Provisioning a pool member is subject to the available pool compute and storage capacity.

Supported Pool Shapes

Pool storage	Available pool compute sizes
256 TB	256 ECPUs
512 TB	256 ECPUs, or 512 ECPUs
1024 TB	256 ECPUs, 512 ECPUs, or 1024 ECPUs

How a Dedicated Elastic Pool Changes Management

Billing and Storage Allocation

Oracle bills each dedicated elastic pool hourly for provisioned storage, in addition to compute billing. Billing starts when you create the pool and ends when you terminate it.

Pool members continue to have their own storage allocation. However, unlike traditional elastic pools, pool members are not billed for storage individually.

For example, in a dedicated elastic pool with 256 TB and 256 ECPUs, the pool leader is billed hourly for both storage and compute. See About Autonomous AI Database Dedicated Elastic Pool Billing and About Billing Autonomous AI Database Dedicated Elastic Pools with Autonomous Data Guard for details.

Maintenance and Patching

When creating or enabling a dedicated elastic pool, you can select a preferred maintenance window, including the day and time. The selected maintenance window applies to all patchable components in the pool.

Only the pool leader can pause or resume patching.

Important:

Pausing or resuming maintenance for a dedicated elastic pool also pauses or resumes maintenance for all other dedicated elastic pools in the same region.
Except for security patches, each dedicated elastic pool must receive a patch at least once every 30 days (4 weeks). Maintenance cannot be deferred beyond this period.
You cannot modify the pool storage (TB) while patching is paused.

If a dedicated elastic pool already exists in a region of your tenancy, additional dedicated elastic pools created in the same region and tenancy cannot have custom maintenance windows. Instead, they inherit the maintenance window of the first dedicated elastic pool created in that region and tenancy. This restriction does not apply when you create a dedicated elastic pool in a different region within the tenancy or in a different tenancy within the same region. Except for security patches, you can pause and resume maintenance multiple times. You can resume previously paused patching at any time from the OCI Console. After patching is resumed, subsequent patching occurs during the scheduled maintenance window.

Autonomous Data Guard

You can enable Autonomous Data Guard on a dedicated elastic pool when another dedicated elastic pool is available:

In the same region, for a local Autonomous Data Guard peer.
In a remote region, for a cross region Autonomous Data Guard peer.

Scaling Pool Capacity

Unlike traditional elastic pools, in which the pool leader can change only the pool ECPU count, dedicated elastic pools allow you to change pool storage (TB) together with pool compute size (ECPUs).

When changing Pool storage (TB):

When scaling down, the new storage value cannot be greater than the storage already used by the pool members.
The new storage value must be within the tenancy limits.
Patching must not be paused.

Database Provisioning

When you create a dedicated elastic pool while provisioning or cloning an Autonomous AI Database instance, the database instance remains in the provisioning state until the dedicated elastic pool is assigned a dedicated server.

Create or Convert a Dedicated Elastic Pool

A pool leader can perform the following operations:

Enable a dedicated elastic pool: Convert an existing elastic pool to a dedicated elastic pool. All pool members move to dedicated infrastructure during the next maintenance window. See Enable a Dedicated Elastic Pool.
Create a dedicated elastic pool while provisioning or cloning an Autonomous AI Database instance: See Create a Dedicated Elastic Pool While Provisioning or Cloning an Instance for instructions.
Create a dedicated elastic pool using an existing Autonomous AI Database instance: See Create a Dedicated Elastic Pool for instructions.

Manage a Dedicated Elastic Pool