Extreme database performance—microsecond latency, petabyte scalability, intelligent block storage—is a strategic advantage for leading global enterprises. But how can smaller organizations get these same capabilities?
With its new cloud-based Exadata Exascale, Oracle bridges that gap. The Exascale intelligent data architecture combines the performance and intelligence of Oracle Exadata, known as the world’s fastest Oracle Database machine, with the advantages of the cloud. Exadata customers of any size can now benefit from multitenant resource pooling and elastic compute and storage, paying for only the resources they use.
This means that smaller firms can now tap into the same high performance database system used by leading Fortune 100 companies, including telcos, chip manufacturers, and financial institutions.
“Exadata Exascale gives a small business the same performance, reliability, and availability that Oracle provides to a stock exchange,” said Kothanda Umamageswaran, senior vice president of Exadata and scale-out technologies at Oracle, during a briefing.
And Exadata Exascale’s multitenant, elastic cloud model opens the door to a broad range of use cases, including departmental workloads that may not need dedicated Exascale servers. Potential scenarios include online transaction processing (OLTP), analytics, and artificial intelligence.
For many Oracle customers, Exadata needs no introduction. When it was first launched in 2008, Exadata offered a powerful combination of software and hardware optimized for peak performance. Then and now, Exadata gets regular updates. Recent state-of-the-art enhancements include faster decryption and decompression and big step-ups in compute cores, memory, and storage capacity to power AI workloads. Exadata also supports complex queries via Oracle’s Smart Scan and AI Smart Scan technologies that offload data-intensive SQL operations from database servers to storage servers, significantly improving query performance and accelerating vector searches.
Exadata Exascale takes these and other leading-edge capabilities and makes them perform better via a loosely coupled cloud architecture. That design removes latency-inducing dependencies that exist in traditional multitier system architectures. Virtual machines, load balancing, and storage metadata have been rethought, removing each as a potential performance bottleneck. Exadata hardware supports Remote Direct Memory Access (RDMA) for direct data transfer from storage to memory, improving latency and throughput.
The value proposition is that Exadata Exascale merges the performance of Exadata’s highly engineered system with the benefits of the cloud. Performance comes from capabilities such as RDMA for fast OLTP input/output (I/O) and automatically offloading SQL queries or vector processing to Exascale’s storage cloud.
Oracle has benchmarked Exadata Exascale I/O latency at 17 microseconds—50 times faster than comparable services from AWS or Azure.
Another feature of the new architecture is database cloning. Exascale can create either full copies or thin clones of a production database or a standby/backup database. The term “thin clone” refers to a clone that populates as needed based on changes to the source database. This can be a money saver because thin clones don’t use as much storage as fully populated production databases. It’s a more efficient model for development and testing.
Exadata Database Service is also available on dedicated infrastructure in Oracle Cloud Infrastructure (OCI), Microsoft Azure, Google Cloud, and AWS, as well as in customers’ own data centers with Exadata Cloud@Customer. In these deployments, customers, or tenants, use dedicated compute and storage servers along with Oracle Automatic Storage Management (ASM), which distributes the storage across databases.
Here’s how Exadata Exascale stands out. It employs a loosely coupled architecture and a shared pool of compute and storage resources in the cloud. So, even if a customer has signed up for just a few compute cores, the Exascale control plane makes hundreds of CPUs available for database queries. That can translate into efficiency, performance, and lower cost.
Another advantage of the loosely coupled architecture is extremely low latency. Oracle has benchmarked Exadata Exascale I/O latency at 17 microseconds—50 times faster than comparable services from AWS or Azure.
The timing is right for Exadata Exascale because more organizations are moving from AI pilot projects to production rollouts. The intelligent data architecture works seamlessly with Oracle AI Vector Search, an Oracle Database 23ai capability for enabling similarity search with both structured and unstructured data as well as retrieval-augmented generation (RAG) with large language models. Exadata Exascale can also accelerate vector searches by using Oracle AI Smart Scan, a unique way of offloading data and compute-intensive AI vector search operations to the Exascale intelligent storage cloud.
Consider a popular “top K” search, where a user may be looking for the top 10 homes or vehicles to match their criteria. Oracle benchmarking shows that, by distributing the search across the Exascale storage cloud and then merging the results, AI vector queries can be executed up to 32 times faster compared with the previous X10M platform. That performance boost really comes into play when thousands of vector searches are concurrently underway.
There’s a direct line from Oracle Exadata Exascale’s shared-resource efficiencies to cost savings, and customers can get started for hundreds of dollars per month. What’s more, Oracle doesn’t charge for different performance levels as measured in I/O operations per second, or IOPS, nor for the total number of I/Os performed.
“We have so much performance, we just say IOPS are included,” said Umamageswaran.
A lot of sophisticated engineering is packed into Exadata Exascale: direct I/O, automatic columnarization, storage buckets, mapping tables, query parallelization, and more. I think of it as years of advanced database engineering delivered as a service.
What’s next? Oracle has signaled that Exadata Exascale is expected to become available through additional offerings in the future, including in multicloud environments. That would be a logical and exciting next step leveraging Oracle’s game-changing multicloud partnerships with AWS, Google Cloud, and Microsoft Azure.
Umamageswaran describes Exadata Exascale as “the future architecture for all Oracle Database Cloud Services.” All the more reason to pay close attention to its evolution.
Oracle Globally Distributed Exadata Database on Exascale Infrastructure is currently in limited availability. The content is intended to outline Oracle’s general product direction. It is intended for information purposes only and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or functionality, and it should not be relied upon in making purchasing decisions. The development, release, timing, and pricing of any features or functionality described for Oracle’s products may change and remains at the sole discretion of Oracle Corporation.
John Foley is editor of the Cloud Database Report and a vice president with Method Communications.
Multicloud deployments offer many benefits, but managing data can be complex. This new ebook provides practical strategies and best practices for connecting data and applications to optimize decision-making.
The future of data analysis is AI-driven. Here’s how to ease integration and access advanced database features and functionality, says John Foley.