By now, most Oracle users are aware of Oracle’s Exadata architecture; how it’s designed to optimize Oracle Database and how it enables key features in the cloud such as Oracle Autonomous Database. They may also be aware that this technology, managed as an Oracle Cloud Infrastructure (OCI) asset, can be deployed in their own datacenter with Exadata Cloud@Customer. What they may not know, however, is that this technology has been undergoing continual improvement for years, and in fairly large, not small increments.

The latest iteration of this product is Oracle Exadata Cloud@Customer X10M, which features improvements over the prior version in many respects, including compute power and capacity, memory optimization, data volume, and overall efficiency, all delivered without a boost in price.

With so many improvements, let’s just consider some salient points.

The prior version featured Intel Xeon processors in both the database nodes and the intelligent storage servers, but with X10M Oracle is using 4th Gen AMD EPYC processors throughout the system. These processors feature up to 96 cores per socket, and organizations can use up to 190 cores on each of their database servers and 64 cores on each of their storage servers. Users now have 3 times the number of database server cores available to process concurrent database queries and, since each core delivers up to 20% more throughput than its predecessor, up to 3.6 times the analytic throughput, according to Oracle. Each 4th Gen AMD EPYC processor is equipped with 12 memory channels to provide superior memory bandwidth and paired with DDR5 memory to deliver faster memory operations. The 4th Gen AMD EPYC processors consume less power per core than previous generation systems to reduce the cost of running the systems and cooling them.

Oracle Exadata compute nodes exchange data with the storage nodes using remote direct memory access (RDMA) to bypass conventional network-based data transfers for faster data movement. RDMA enables the sender to put data directly in the receiver’s memory and retrieve data from that memory without involving the receiver’s processors at all. The speed of that process is further enhanced by a RDMA over Converged Ethernet (RoCE) internal fabric that enables an aggregate active-active bandwidth of 200 gigabits per second between servers. The data is tiered in the storage server using the RDMA-enabled memory, called Exadata RDMA Memory (XRMEM), NVMe PCIe Flash, and high-capacity storage drives. According to Oracle, the result is that SQL Read Latency is less than 17 microseconds, which can reduce transaction processing latency and improve its throughput.

Exadata also lets users run multiple VM Clusters across database servers, which enable multiple independent instances of Oracle Database to run on a single Exadata system, simultaneously providing infrastructure consolidation with operational isolation. It gives better control over the use of very different database workloads that share the same hardware, and allows, among other things, for Oracle Autonomous Database and the regular Exadata Database Service on the same Exadata Cloud@Customer X10M infrastructure.

For management purposes, including identity management, billing, and patching, Oracle Exadata Cloud@Customer is connected to an Oracle Public Cloud Region or Dedicated Region. Oracle manages the Cloud@Customer platform remotely while customers have control over their data throughout its life cycle, enabling them to meet data residency requirements if needed. Customer data is always encrypted on Exadata Cloud@Customer and Oracle does not have access to the encryption keys so it cannot access private data when remotely managing the system. In addition, Oracle Autonomous Database and Exadata Database Service running on Exadata Cloud@Customer X10M are the same as the versions running in OCI. This means that databases, scripting, applications, and data management expertise can easily move from on-premises datacenters to Oracle’s cloud and back with minimal effort. This capability makes it easy for users to start using OCI database services locally and then transfer entire workloads to the cloud when they’re ready.

What does all this mean? Current users of Exadata Cloud@Customer that upgrade to X10M will enjoy an immediate performance boost at no additional cost, and Oracle Database users that are not availing themselves of Exadata capabilities can expect to see their database workloads run faster and on less total infrastructure if they consolidate operations on the new platform. This platform using 4th Gen AMD EPYC processors is, according to Oracle, the future direction of Exadata, so one can expect that Exadata in OCI will soon also take advantage of this technology. In addition, Microsoft Azure and Oracle Database users will benefit from this higher performance platform as Oracle incorporates Exadata on OCI into Azure datacenters with Oracle Database@Azure.

IDC’s point of view: Oracle has long seen its mission as delivering the fastest, most scalable, and most robust database system in the industry. To this end, the company developed Exadata, which is meant to overcome limitations of commodity hardware and deliver capabilities specifically designed for Oracle Database operation.  It has steadily enhanced this platform over more than a dozen years. Oracle Exadata is designed specifically to optimize the operation of Oracle Database, so it is quite naturally the best platform for this RDBMS. The latest enhancements to the core Exadata infrastructure, including AMD EPYC processors, signal a strong commitment on the part of Oracle to continue delivering the best possible user experience to its customers.

Finally, some future considerations. Exadata X10M is designed to support large data workloads with concentrated compute power. As users move to take advantage of such multi-model features as the JSON Relational Duality, they will find the system is up to the task. In addition, Oracle’s support for AI vector development and management will require processing large amounts of data, but again, that should not be a problem for Exadata X10M. In short, X10M is another important step for users in addressing a world requiring rapidly increasing data volumes and compute performance.

In summary:

  • Oracle is rolling out a powerful new version of Oracle Exadata called X10M.
  • This version features faster, denser, and more energy efficient processing thanks to the use of 4th generation AMD EPYC processors.
  • Enhancements in memory capacity and internal data transfer speeds mean that Oracle Database on Exadata X10M, including Oracle Autonomous Database, can process more data faster than on the previous X9M platform, yet without an increase in cost to the user.
  • Users of Oracle Database that demand fast performance for large, complex workloads should take a hard look at this system.
  • IDC believes that many enterprises will find themselves in a bind, dealing with rapidly growing data volumes and performance demands. With hundreds of deployments in mission-critical applications Exadata has demonstrated that it can address these challenges successfully. Exadata X10M enhances Oracle’s opportunity for continued success.

 

 

For more information on Exadata Cloud@Customer X10M, visit Oracle’s web site at https://www.oracle.com/engineered-systems/exadata/cloud-at-customer/