Oracle has unveiled its latest database offering it claims will enhance agentic AI deployments and help with data sovereignty demands.

The Oracle Globally Distributed Exadata Database on Exascale Infrastructure lets users deploy applications across disparate servers, which the hyperscaler claims acts as an always-on architecture.

Oracle’s new service distributes, stores, and synchronizes data across multiple locations, meaning users looking to keep ahead of the newfound desire for data sovereignty can use it to keep in line with compliance requirements.

The firm also claims the new database service can keep applications running in the event of regional outages. The distributed database uses raft-based replication to provide an active architecture across disparate data centers. This means that in cases where a facility is taken offline, like in the recent Russia earthquakes, the system simply fails over to another.

Speaking to SDxCentral ahead of the launch, Wei Hu, SVP of high availability technologies at Oracle, said the resiliency aspect was already being eyed by several users.

“Because things are online now, everybody's very sensitive to outages. Banks and credit card companies are interested in this. Even infrastructure companies, with one example being a European firm with their smart meter codes, they want real-time data so they can adjust to changes in demand and so forth.”

Data residency boon for agentic AI

The database offering’s ability to scale up or down to meet the needs of changing workloads comes as one emerging technology is causing increasing fluctuations on networks: AI agents.

For all the excitement around agentic AI, its impact is set to cause headaches for network operators, with individual agents running variable workloads simultaneously, causing intense fluctuations on backend systems as they perform tasks like vector searches. An executive from Cisco recently suggested to SDxCentral that AI agent deployments could soon strain network bandwidths equivalent to “80 billion” users.

Oracle touts its new distributed exascale database as a way to prevent that strain to better provision and manage agentic AI deployments, building on the groundwork laid by its Globally Distributed Autonomous Database, released last March.

Its new database offering, however, manages that provisioning, while ensuring that data being used by those AI agents is kept within the regions where they’re needed.

With AI data placed in one distributed database, instead of multiple single-purpose databases, Oracle argues its offering can help reduce strain on backend systems while keeping related data compliant with local residency laws, like India’s Digital Personal Data Protection Act (DPDP) or the GDPR or Data Act in the EU.

“The increasing prevalence of agentic AI [and] the increasing need for data residency requirements in specific locations – keeping that AI data in that specific location – the confluence of the two, we think this is designed to really solve that,” Steve Zivanic, Oracle’s global VP for database and autonomous services, said.

“And with the serverless architecture, basically any organization can subscribe to this in any location.”

Serverless scaling

Beyond data sovereignty, Oracle’s new distributed Exadata database comes as network operators are looking for increased resilience.

The hyperscaler suggests its offering can help address the packet drops and data loss failover, with its Exascale platform’s fault-tolerant architecture.

Instead of replicating the entire application stack in every country where it’s deployed, Oracle’s database solution creates writable clones of the entire database without ETL (Extract, Transform, Load) cloning – meaning users can create snapshots of their data almost instantly from any location without consuming additional storage space

“When I add a new server, I don't have to copy data anymore,” Hu explained. “I just point to the existing storage, and it immediately starts serving the device. That means now when users add a service, it near instantaneously copies data, and the storage itself can be scaled very quickly, from terabytes to exabytes.”

With Broadcom’s VMware issues causing customers to turn their heads for alternatives, Oracle touted the fact that users can leverage its new database offering across multiple clouds.

“You can deploy data across multiple locations by splitting a single database and storing its parts in different places,” Hu outlined.” So you can store part of your data in AWS’s product cloud, or Google Cloud, or Azure.

“What’s useful is that if one provider hikes its prices on you, you can start shifting your data out to a different cloud, and the application doesn’t change. This elasticity is actually very important from a business perspective.”