Guest post by Carl Olofson, Principal Analyst, DBMSGuru LLC
Oracle has rolled out a configuration of its flagship product, Oracle Database, in a form that provides automated support for a single logical database that spans regions without requiring the labor-intensive and error-prone activities involved in attempting to span a database across regions manually.
A common approach, where enterprises want to run a critical database across regions, is to manually break it into overlapping shards and use replication to synchronize them, usually designating each shard instance as “owner” of a piece of the database. This means that any query on any region will yield results that are partly a little stale when the replication has not finished the sync operation. It also requires special coding of the applications to avoid conflicts. Various vendors have offered globally distributed single logical database functionality, usually syncing updates using some sort of universal clock, but these all have their pitfalls. Also, to use them, Oracle users, who have a considerable investment in the Oracle Database as a data platform, would be forced to re-write applications as they convert to systems that are somewhat incompatible with the rest of their database and application environment.
Oracle has an elegant extension to Oracle Database, called Oracle Globally Distributed Database (GDD), which distributes the database transparently to the desired regions and uses internal functions to synchronize the data and optimize operations. GDD is available as built-in functionality to all Oracle Database Enterprise Edition users – in OCI, multicloud, and on-premises environments.
In addition, Oracle has now introduced Globally Distributed Exadata Database on Exascale Infrastructure (GDD on Exascale) as a cloud service on OCI. This new offering uses an intelligent data architecture with Exadata optimizations and OCI automation to support nearly limitless transparent and dynamic scalability for existing applications or new workloads.
Like other geo-distributed database approaches, GDD on Exascale deploys the data in what Oracle calls shards, but unlike shards that are managed manually, these are connected in a data sharing network that allows the data to be physically partitioned in a manner that is automatically optimized and supports continuously concurrent operations across shards. Each shard is replicated for recoverability. Transactions are distributed automatically, and their consistency maintained by a technology in Oracle Database 23ai called Raft replication, which obviates the need for a universal clock. Query processing takes advantage of the networked parallelism to enable massively parallel analytics and AI vector processing.
Oracle introduced the essential elements of distributed database functionality back in 2017, and that technology already powers a number of globally deployed critical applications in areas that include mobile messaging, credit card fraud detection, payment processing, internet infrastructure data management, smart power meters, and data sovereignty.
Oracle points out that as users increasingly deploy agentic AI workflows, demands on the database for both AI vector searches and backend transaction processing are exploding and require a “hyperscale serverless database system” that scales up or down on demand to meet spikey workload demands. GDD on Exascale also delivers the ability to cleanly and consistently support data residency, helping customers address requirements that access data governed by rules present within the jurisdictions in question.
Oracle represents a summary of the key elements of this offering as follows:
- Fully featured, totally transparent, enterprise-grade distributed relational database support
- As always, as a multi-functional DBMS, optimized support for OLTP, petabyte scale analytics, and agentic AI.
- Built-in native SQL, supported by internal architecture, and designed to optimize performance while avoiding potential data loss or inconsistencies
- Six available partitioning and distribution methods provide automatic support for strategically collocated data, helping provide low latency, regulatory compliance, and overall efficiency
- Three forms of replication providing support for the variety of network conditions that the user may have to face
- Exascale’s serverless architecture reduces cost and provides dynamic elasticity
For smaller organizations, GDD on Exascale provides serverless scalability (that is, it can be scaled down to zero processors if needed) for a low cost entry point while improving optimal parallel query performance. For large, global organizations, Oracle’s new GDD service enables distributed database functionality with the operational elements built in. Oracle customers can look to GDD on Exascale to provide a cost-effective, transparently scalable, global, and automatically managed environment for their mission-critical databases.