Why Oracle 23c’s Global Distributed Database Is a Game-Changer for Multi‑Region Data

Oracle has long been a flagship of centralized databases, with its RAC technology standing unrivaled; consequently it entered the distributed‑database arena relatively late. After 2010, the rise of internet companies spurred distributed‑database technologies, and the emergence of NVMe SSDs and share‑nothing architectures accelerated the trend, prompting Oracle to introduce sharding in 12c.

Initially I assumed Oracle’s sharding was meant to boost OLAP capabilities in OLTP databases, but during the 2015 CAB "ASK ANDY" session Andy clarified that Oracle sharding targets OLTP scenarios, especially IoT, not OLAP. Early Oracle Sharding was limited, relying on manually deployed ADG for high availability and lacking clear use cases, so it quickly faded from attention.

When designing an Oracle database appliance later, I leveraged NVMe SSDs and experimented with cross‑instance parallel queries on a four‑node RAC. I discovered that, provided storage performance and CPU resources are sufficient, Oracle’s cross‑instance parallel query can match or even surpass many distributed databases, thanks to its sophisticated optimizer, reducing my interest in Oracle’s distributed‑database offerings.

Oracle’s sharding remained a modest feature in subsequent releases, mentioned but not promoted, until Oracle 23c introduced a notable change. Early previews revealed that 23c supports RAFT replication groups, confirming that Oracle is moving toward a genuine distributed database beyond basic sharding, potentially combining both distributed and RAC architectures.

However, a closer look shows the strategy is framed around a new name: Oracle 23c "Global Distributed Database". The term "global" was initially misread as "global" in the generic sense, but Oracle confirms it refers to a truly global distributed database.

Oracle 23c’s Global Distributed Database enables data storage and processing across multiple geographic locations, supporting cross‑shard, cross‑cluster, and cross‑cloud data management and queries. Its primary use cases involve cross‑region or even cross‑country deployments.

This solution is not aimed at scaling within a single availability zone or data center—RAC already addresses intra‑data‑center horizontal scaling. Instead, it targets scenarios that span data centers worldwide, allowing instances to be deployed in different regions or clouds while maintaining data consistency and availability. Typical applications include internet finance, e‑commerce, gaming, large‑scale social networks, IoT, and AI.

A key driver behind Oracle’s global offering is compliance with data‑localization regulations. Many countries now require that citizen data remain within national borders. Oracle 23c’s Global Distributed Database can create partitions per country, keeping local data on‑premises while still supporting global queries. Its TrueCache feature caches remote data locally in memory without persisting it, satisfying regulations that forbid cross‑border persistent storage.

Discussions with peers from domestic database vendors confirm that Oracle’s approach reflects a deep understanding of user requirements, turning those insights into a product that addresses both performance and regulatory challenges.