Oracle recently launched of its latest Exadata platform, the Oracle Exadata X11M. Purpose-built to optimize the performance of Oracle Database, including Oracle Autonomous Database, X11M is a significant leap forward in speed and efficiency compared to previous generation Exadata systems for cloud, multi-cloud, and customer data centers.

The X11M’s design is the product of Oracle’s deep expertise in database development, with deep engineering between the database software and the hardware optimized to support it.

Exadata X11M

The Oracle Exadata X11M is the latest iteration in Oracle’s Exadata platform, engineered for high-performance Oracle Database workloads across AI, OLTP, and analytics. This twelfth-generation Exadata system integrates advanced hardware and software technologies to deliver unparalleled throughput, scalability, and efficiency.

Below is a detailed breakdown of its key technical features.

Compute Architecture

1. Processors: X11M utilizes AMD EPYC 9004 series processors with up to 96 cores per CPU. These processors offer a 25% increase in per-core performance compared to the X10M generation, making them ideal for compute-intensive workloads.
2. Memory: The platform supports DDR5 memory, with configurations providing up to 33% more bandwidth over DDR4. This enhancement improves memory throughput, which is critical for large-scale analytics and AI workloads.
3. Database Server Configurations:
   1. Up to 1,390 GB of usable memory per database server in public cloud configurations.
   2. On-premises configurations offer flexible memory tiers, scaling to 3 TB per server.

Storage System Enhancements

1. High Capacity and Extreme Flash Options:
   • High Capacity (HC) servers: Feature 12 x 22TB hard drives with a combined raw capacity of 264TB per server.
   • Extreme Flash (EF) servers: Use NVMe SSDs delivering 122.88TB of raw flash capacity.
2. Performance:
   • Flash throughput has been increased to 100GB/s per storage server.
   • Combined flash and RDMA memory throughput for analytic scans reaches 500GB/s per server, a 2.2x improvement over X10M.
   • RDMA memory read latency has been reduced to 14 microseconds, a 21% improvement.

Networking Infrastructure

1. RDMA Over Converged Ethernet (RoCE):
   • Dual-port RoCE-based network fabric with PCIe Gen 5 interconnects ensures ultra-low latency for data transfers.
   • RDMA access between compute and storage minimizes CPU overhead, improving transaction and analytics processing efficiency.

Enhanced AI Capabilities

Oracle’s Exadata X-series database machines were initially designed for analytics and OLTP workloads, where the platform continues to set the bar for performance and reliability. Over the past several years, however, AI has grown to equal prominence. The X11M and the new AI-targeted capabilities in recent Oracle Database releases recognize this expansion.

The X11M accelerates AI workloads by offloading complex vector processing to storage servers, reducing data transfers and improving efficiency. AI queries are up to 32 times faster using binary vector formats, while optimized vector distance functions reduce CPU usage and improve query performance by up to 4.6 times.

AI workloads using hierarchical navigable small-world, HNSW, indexes see performance improvements of up to 43%, while inverted file flat indexing on storage servers delivers up to 55% faster query execution. These are remarkable numbers.

OLTP and Analytics Updates

Oracle isn’t ignoring the needs of its traditional OLTP and analytics customers. The Exadata X11M supports up to 1.25 times more concurrent transactions, up to 21 percent lower SQL 8K I/O read latency — now only 14us, achieving up to 1,000,000 write IOPS on storage servers to increase the throughput for transactional workloads.

For analytics, Exadata X11M uses faster processor cores, memory, and flash in storage servers to increase overall analytic query performance by 25%. Enhanced Exadata System Software also enables Oracle’s Smart Scan technology to scan data in both flash and RDMA memory, enabling analytic SQL throughput of up to 500 GB/s per storage server and multiple TB/s in even small configurations.

Sustainability and Efficiency

Sustainability is top-of-mind for nearly every enterprise, which Oracle addresses in the X11M with its new intelligent power-saving features. These features can cap CPU power consumption, adjust CPU usage based on workload demands, and reduce energy consumption during periods of low activity. Unused processor cores can also be turned off to conserve power, lower operational costs, and enhance sustainability.

Exadata X11M’s extreme performance levels also allow customers to consolidate workloads further, reducing their hardware footprint and operational overhead, enabling organizations to achieve even greater cost and environmental efficiencies.

Day One Cloud Integration

Oracle’s strategy for deploying Exadata has shifted over the past several years to keep up with the evolving requirements of its enterprise customers, who increasingly rely on cloud services for business-critical applications.

Oracle initially brought Exadata to its own Oracle Cloud Infrastructure (OCI). More recently, they surprised many industry watchers by making the platform natively available on OCI within the data centers of its competitors, Microsoft Azure, Google Cloud, and Amazon Web Services.

The new X11M solution is available at launch across all of its deployment options: public cloud partners, its own Oracle Cloud Infrastructure, its Oracle Cloud@Customer as-a-Service offering, and traditional on-prem deployments.

Analysis

The Exadata X11M builds on Oracle’s engineering expertise to address the evolving needs of database customers. It provides a robust platform that supports high-demand workloads such as artificial intelligence, online transaction processing, and analytics.  

Key benefits of the new offering over the previous generation Exadata include:

• Performance Improvements: Data-optimized hardware components and data-intelligent software enhancements allow quicker transaction processing, accelerated analytics, and AI workloads that can see up to a 30 times speedup, according to Oracle.
• Scalability: Exadata can be configured to match current operational needs and scaled elastically as those needs grow, ensuring customers can optimize their infrastructure investments over time.
• Cost Efficiency: Customers can save by consolidating more workloads onto fewer systems. New power-saving features and optimized resource usage contribute to reduced operational costs.

Being available on leading public cloud providers at launch lets the X11M meet customer demands where they operate. Enterprises can now leverage Oracle Database and Exadata capabilities within their chosen cloud ecosystems, optimizing costs and simplifying IT infrastructure.

The flexibility to run Exadata in multi-cloud configurations enables organizations to balance workload demands, manage data sovereignty requirements, utilize existing cloud service commitments, and access data in Oracle databases more easily from their chosen cloud applications. This powerful approach makes the Exadata X11M affordable to nearly every enterprise and Oracle Database customer.

The Oracle Exadata X11M is a significant step forward in database technology, delivering performance and scalability that meet the demands of modern enterprises. It offers customers flexibility and operational efficiency while supporting the continuing strategic shift by enterprises toward hybrid and multi-cloud architectures. For database customers seeking to optimize workloads and future-proof their operations, the Exadata X11M presents the perfect solution.