Oracle recently announced that it’s expanded its Oracle Database@Azure service with the general availability of Exadata Database Service on Exascale Infrastructure, introducing a shared, multi-tenant deployment model. The service now spans 14 Azure regions, with 18 additional regions planned within the next 12 months.
Oracle also announced the upcoming availability of Base Database Service on the platform and added support for the next-generation Exadata X11M platform on dedicated infrastructure.
This expansion represents Oracle’s continued multi-cloud strategy, particularly its deepening partnership with Microsoft. The Exascale offering targets organizations seeking Exadata capabilities without the traditional infrastructure commitments, though actual cost benefits will depend heavily on workload characteristics and usage patterns.
What was Announced?
The Exascale Infrastructure introduces several architectural changes to Oracle’s traditional Exadata model. Oracle positions this as a “loosely coupled, hyper-elastic architecture” that maintains Exadata’s core capabilities while enabling more granular resource consumption.
Key Technical Capabilities
The Exascale deployment model implements elastic, pay-per-use resources where customers specify database server ECPUs and storage capacity without additional IOPS charges.
The minimum configuration consists of two virtual machines, each containing 8 ECPUs and 22 GB of memory, plus 300 GB of intelligent database storage. The system supports scaling from this baseline to 100 TB of storage capacity.
Oracle’s intelligent storage cloud distributes databases across pooled storage servers using RDMA-capable infrastructure. Oracle states that this architecture can deliver up to 2,880 GB/s of throughput for analytics workloads through automatic query offloading to storage-layer processing. For AI workloads, AI Smart Scan can accelerate vector search operations by up to 30X compared to traditional approaches.
Platform Compatibility and Migration
The platform exclusively supports Oracle Database 23ai, incorporating features such as AI Vector Search and JSON Relational Duality. This version requirement ensures access to Oracle’s latest database capabilities but creates potential migration complexity for organizations running older database versions.
Migration utilities include Oracle Data Guard for physical database replication, GoldenGate for logical replication, and Data Pump for data export/import operations. The vendor claims full architectural compatibility with on-premises Exadata environments, enabling lift-and-shift migrations without application modifications.
Hardware Platform Updates
Oracle Exadata Database Service on Dedicated Infrastructure now supports its recently launched Exadata X11M platform, which Oracle claims delivers performance improvements across AI, analytics, and OLTP workloads compared to previous generations.
The X11M platform is the latest iteration of Oracle’s engineered systems approach, though specific performance metrics, architectural enhancements, and deployment timelines were not disclosed in the announcement.
The X11M platform maintains compatibility across on-premises, hybrid cloud, public cloud, and multicloud deployment models, enabling consistent performance characteristics regardless of deployment location. This consistency aims to simplify hybrid cloud architectures and reduce performance variability across different deployment environments.
Impact to IT & Database Teams
The Exascale model addresses a longstanding barrier to Exadata adoption by eliminating the need for large upfront infrastructure commitments. Traditional Exadata deployments required significant minimum configurations that often exceeded the immediate needs of smaller organizations. The pay-per-use model with granular scaling enables a more predictable cost management for variable workloads.
Database administrators benefit from maintaining compatibility with on-premises Exadata environments, thereby reducing the need for architectural changes during cloud migration.
The automated lifecycle management and rapid cloning capabilities may reduce administrative overhead, particularly for development and testing environments.
Its pay-per-use model benefits workloads with variable demand patterns but may prove more expensive than dedicated infrastructure for consistently high-utilization scenarios. Organizations should model their specific usage patterns against both deployment options.
The absence of IOPS charges offers cost predictability for I/O-intensive workloads; however, the overall pricing structure’s competitiveness depends on workload characteristics and the regional availability of alternative solutions.
Analysis
Oracle’s expansion of Database@Azure with Exascale Infrastructure is a forward-looking response to the enterprise demands for more flexible database infrastructure consumption models. The service addresses legitimate barriers to Exadata adoption while maintaining Oracle’s architectural differentiation through the integration of engineered systems.
Oracle’s approach differentiates through a deep integration of database and infrastructure optimization. The ability to offload processing to intelligent storage layers gives Oracle an architectural advantage that competitors using standard cloud infrastructure can’t easily replicate. The RDMA-enabled, database-aware storage architecture provides technical differentiation in performance-critical scenarios.
The multicloud strategy, particularly the Azure partnership, addresses enterprise requirements for avoiding single-cloud dependencies. Running Oracle Database on Azure within Oracle Cloud Infrastructure within Azure data centers provides technical integration benefits while maintaining vendor relationship flexibility.
Organizations with significant Oracle Database investments and requirements for high-performance analytical or AI workloads represent the most logical early adopters.
Competitive Outlook & Advice to IT Buyers
These sections are only available to NAND Research clients. Please reach out to [email protected] to learn more.
