Overview of Oracle Globally Distributed Autonomous AI Database

Learn about the key capabilities of and the concepts you need to know to use Oracle Cloud Infrastructure Globally Distributed Autonomous AI Database service.

About Oracle Globally Distributed Autonomous AI Database

Globally Distributed Autonomous AI Database brings the power of distributed (sharded) databases to Oracle Autonomous AI Database on Dedicated Exadata Infrastructure.

Oracle Globally Distributed Autonomous AI Database is a cloud-based, fully-managed database service that enables the sharding of data across globally distributed converged databases. It is designed to support large-scale, mission-critical applications. It is a highly available, fault-tolerant, and scalable database service that enables organizations to store and process massive amounts of data with high performance and reliability.

The Globally Distributed Autonomous AI Database is built on top of Oracle’s autonomous technology, which means that it is self-driving, self-securing, and self-healing. This allows automation of many of the routine tasks associated with managing a database, such as patching, tuning, and backup and recovery, which can help reduce the risk of human error and improve system uptime.

For a detailed discussion of distributed database features supported, see Oracle Sharding Overview for Oracle Database 19c and Oracle Globally Distributed Database Overview for Oracle AI Database 26ai.

Globally Distributed Database Concepts

To gain a greater understanding of Globally Distributed Database concepts, familiarize yourself with the following terminology.

• Catalog: a database that supports automated shard deployment, centralized management of the distributed database, and multi-shard queries.

  A Catalog serves following purposes:

  • Serves as an administrative server for the entire distributed database

  • Stores a gold copy of the database schema

  • Manages multi-shard queries with a multi-shard query coordinator

  • Stores a gold copy of duplicated table data

• Shard: A distributed database is a collection of shards.

  Each shard in a distributed database is an independent database instance that hosts subset of the distributed database data. Shared storage is not required across the shards.

  Shards can all be placed in one region or can be placed in different regions.

  Shards are replicated for high availability and disaster recovery with Raft replication or Oracle Data Guard. For high availability, Raft replicated shards can be placed in different availability domains within a region. Data Guard standby shards can be placed in the same region where the primary shards are placed for high availability. For disaster recovery, the Data Guard standby shards can be located in another region.

• Shardspace: A shardspace is a shard that stores data corresponding to a range or list of key values in a user-managed data distribution configuration. A shardspace consists of a shard and its replica.

• Shard director: A network listener that enable high performance connection routing based on a sharding key.

  In addition, a shard director is a set of processes known collectively as a Global Service Manager (GSM) that acts as a regional listener for clients that connect to a Globally Distributed Database.

  The shard director maintains a current topology map of the distributed database. Based on the sharding key passed during a connection request, the director routes the connections to the appropriate shard.

• Global service: A database service that is used to access data in the distributed database.

  A global service is an extension to the notion of the traditional database service. All of the properties of traditional database services are supported for global services.

For more in depth information about distributed database components and schema objects see Architecture and Concepts in Oracle Globally Distributed AI Database Guide.

Data Replication Solutions

Oracle’s Globally Distributed Autonomous AI Database service offers data replication solutions to ensure high availability, disaster recovery, and additional scalability for reads.

Globally Distributed Autonomous AI Database offers shard-level replication with Oracle Data Guard on Oracle Database 19c and Oracle AI Database 26ai. Raft replication is available with Oracle AI Database beginning in release 26ai.

Globally Distributed Autonomous AI Database automatically deploys the specified replication topology to the procured systems, and enables data replication.

Shard-Level Replication with Oracle Data Guard

A shard is a database. Oracle Data Guard replication of shards to physical standby databases can be used to provide individual shard-level high availability. Replication is automatically configured and deployed when the distributed database is created.

Oracle Data Guard is tightly integrated with Oracle’s Globally Distributed Autonomous AI Database service to provide high availability and disaster recovery with strict data consistency and zero data loss. Oracle Data Guard replication maintains synchronized copies (standby databases) of shards (the primary databases) for high availability and data protection. Standbys can be deployed locally or remotely.

Chunk Set-Level Replication with Raft Replication

Instead of replication at the whole shard level using additional databases for standbys, the Raft replication feature in Globally Distributed Autonomous AI Database creates sets of chunks of data from each shard and distributes them automatically among the shards to handle chunk assignment, chunk movement, workload distribution, and balancing upon scaling (addition or removal of shards), including planned or unplanned shard availability changes.

Raft replication is built into Globally Distributed Autonomous AI Database to provide a consensus-based, high-performance, low-overhead availability solution, with distributed replicas and fast failover with zero data loss, while automatically maintaining the replication factor if shards fail. With Raft replication management overhead does not increase with the number of shards. If you are used to NoSQL databases and do not expect to know anything about how replication works, native replication just works.

Unlike Data Guard replication, Raft replication does not need to be reconfigured when shards are added or removed, and replicas do not need to be actively managed.

For more details about how Raft replication works see Using Raft Replication in Oracle Globally Distributed Database.

Resource Identifiers

Oracle’s Globally Distributed Database services resources have a unique, Oracle-assigned identifier called an Oracle Cloud ID (OCID).

Globally Distributed Autonomous AI Database resources are listed here.

Resource	Identifier
Distributed Autonomous Database	osddistributedautonomousdb
Distributed Database Private Endpoint	osddistributeddbprivateendpoint
OSD Work Request	osdworkrequest

For example, the OCID format for a Distributed Autonomous Database resource is ocid1.osddistributedautonomousdb.oc1.iad.<UNIQUE ID>.

For information about the OCID format and other ways to identify your resources, see Resource Identifiers.

Metering and Billing

Metering and billing for Globally Distributed Autonomous AI Database is based on the number of ECPU per hour.

Because ECPUs are allocated in the Autonomous AI Database, see Compute Management and Billing for details.

Note: Once you tag a cluster for use in a Globally Distributed Database, it will continue to bill for the Globally Distributed Database SKU until the cluster is deleted.

Service Limits

Globally Distributed Database Service Limits can be set for Distributed Database Count and Distributed Database Private Endpoint Count.

Autonomous AI Database instances, ECPU count, and storage need to have limits set for Autonomous AI Database service.

See Plan and Monitor Capacity for details.

Integrated Services

Oracle’s Globally Distributed Database services are integrated with various Oracle Cloud Infrastructure services and features.

IAM

Oracle Globally Distributed Database services are integrated with the Identity and Access Management (IAM) service for authentication and authorization for the Console, SDK, CLI, and REST API.

To learn more about IAM, see IAM Overview.

Work Requests

Globally Distributed Autonomous AI Database uses its own APIs for Work Requests.

To monitor work requests see Monitoring Work Requests.

The permissions required for using the APIs are documented in Permissions for Globally Distributed Autonomous AI Database APIs.

Monitoring

Oracle Cloud Infrastructure Monitoring lets you actively and passively monitor your Globally Distributed Database resources and alarms.

Globally Distributed Database metrics capture CPU utilization, OCPU consumption, memory utilization, deployment health, and inbound and outbound lag. You can view these metrics using the Monitoring service.

See Monitoring a Globally Distributed Database for more details about monitoring the health and performance of a distributed database.