Adding Shards

Add shards to scale out your Globally Distributed Autonomous AI Database.

You can add shards when:

• You have completed Creating a Globally Distributed Autonomous AI Database Resource, but have not yet started Deploying Globally Distributed Autonomous AI Database.

• You have completed Deploying Globally Distributed Autonomous AI Database and want to scale up your Globally Distributed Autonomous AI Database with more shards.

1. On the Details page, on the Shards tab, select Add Shard.

2. On the Add Shards panel configure the new shard.

   In Shard Count indicate the number of shards you want to add

   You can edit each shard you add in the table, by selecting Edit in its actions menu (three dots).

   • Shard: The display name for this shard in the configuration.

   • Region: Region where the shard will be added.

   • VM cluster: Select a cluster available in the selected region.

     Note: It is recommended that you use one VM cluster per database (shard or catalog).

   • ECPU count: The number of ECPU cores to enable. Specify the number of ECPUs for your shard as an integer. Available cores are subject to your tenancy’s service limits. You must enter a minimum of 16 ECPUs per shard.

   • Auto scaling: Enable automatic scaling based on workload per shard/shardspace

   • Storage: GB of storage to allocate to your database

   • Enable Data Guard: Instantiates Oracle Data Guard standby instances for each shard

   • Data Guard region: Select the region where you would like to host the shard's Data Guard standby

   • Data Guard VM Cluster: Select a cluster available in the selected Data Guard region.

     Note: You can select a cluster that contains a primary shard for a Data Guard standby database; however, it is recommended that you use one VM cluster per database (shard or catalog).

   You can add up to 10 shards in each set to deploy, and then add more after deployment if needed.

3. In Create administrator credentials, set the password for the shard database ADMIN user.

4. Select the Encryption key details for the new shards.

   Note the following:

   • OCI Vault Service (KMS) should be used if you will have shards in less than 3 regions.

   • If you will have shards in 3 or more regions, Oracle Key Vault should be used for encryption.

   • After the distributed database is created with an encryption key type, you cannot change to a different type, for example, you cannot change a shard from using OCI Vault Service (KMS) to Oracle Key Vault, or the reverse.

   The encryption key settings you configure depend on the data distribution method configured for the Globally Distributed Autonomous AI Database when it was created.

   Automated data distribution:

   • OCI Vault Service: All shards have the same encryption vault and encryption key, and is mandatory.

   • Oracle Key Vault: All shards have the same OKV endpoint group name.

   User managed data distribution:

   • OCI Vault Service: Each shard can have the same or different encryption key details, and is optional.

   • Oracle Key Vault: Each shard can have the same or different OKV endpoint group details, and is optional.

   For both cases:

   • Based on the primary region that you selected for the first shard, you select a key type (OCI Vault service or Oracle Key Vault), and the vaults/keystores and encryption key/OKV endpoint group available in that region and selected compartment.

   • If Data Guard is enabled for a shard, and if the standby region is not the same as the primary region, you can:

     • OCI Vault Service: Select virtual private vaults that are replicated in the standby region.

     • Oracle Key Vault: The keystore and OKV endpoint group of the primary region are used by default.

   • If you use Oracle Key Vault, ensure that the OKV endpoint group is valid and preconfigured according to the system requirements.

5. Select Validate to run checks to make sure the new shards are valid.

6. Once any validation errors are addressed and validation is successful, select Add to finish creating the new shards.

7. Once created, it is a best practice to validate the CA bundles, to verify that the CA bundles are equivalent on any new clusters added to the distributed database. See Validating CA Bundles.

Note: There is a time limit for deploying new shards.

• When scaling up a deployed Globally Distributed Autonomous AI Database, you must complete Deploying Globally Distributed Autonomous AI Database within 7 days of completing this procedure or you will get an error and must terminate the new shard resources and start again.

• When adding shards to an undeployed Globally Distributed Autonomous AI Database, you have 7 days from completing Creating a Globally Distributed Autonomous AI Database Resource to add any shards and complete Deploying Globally Distributed Autonomous AI Database.

For more information about the concepts and considerations of adding shards to a Globally Distributed Autonomous AI Database see Shard Management in Using Oracle Sharding.