Oracle’s Latest Exadata Machines Available As Public Cloud Service

The ultra-fast X8M can now be provisioned, with granular flexibility in resource utilization, massive scalability, and pay-by-the-second billing, on Oracle Cloud Infrastructure


Oracle’s most-advanced Exadata machines, delivering higher network throughput, greater compute and memory scalability and granular resource provisioning, became available Wednesday as an on-demand cloud service.

The Exadata X8M, previously offered only for on-premises deployment, can now be provisioned on Oracle’s public cloud to power challenging workloads such as mission-critical transaction processing systems, analytics, petabyte-scale data warehouses and critical business applications like SAP, Oracle E-Business Suite and Fusion ERP.

“It’s been a very fast ramp for this product in the cloud,” Juan Loaiza, Oracle’s executive vice president for mission-critical database technologies, said in a news conference unveiling the service. “Customers think this is very cool, very leading-edge technology.”

Sponsored post

[Related: Oracle’s Cloud Roadmap Features New High-Power Instances]

Oracle launched the X8M, the latest in a series of hardware-software integrated Exadata servers geared for data-intensive workloads, in September 2019 at its OpenWorld conference.

The public cloud variant delivers on-demand infrastructure, billed by the second, that’s identical to what customers can put in their data centers as either traditional on-premises servers or through the pay-as-you-go Cloud@Customer offering.

The new system is an upgrade over the X8—with the added M standing for memory. But it sells for the same price as its predecessor.

The X8M introduces an RDMA networking fabric that clocks at 100 Gb/sec (the previous generation was 40 Gb/sec) and up to 96 terabytes of persistent memory, delivering a 2.5x increase in application transaction processing IO over the previous Exadata Cloud Service release and 10x improvement in IO latency.

The “big thing is persistent memory in storage,” Loaiza said. “We’ve developed a lot of software to make that work really well.”

The higher network throughput makes the new Exadatas “really good for analytics,” Loaiza added.

Oracle also dramatically ramped the system’s ability to scale to a whopping 4,600 CPU cores supported by 44 TB of DRAM and 1.6 PB of flash storage.

“All these are huge numbers,” Loaiza said. Short of running Facebook, “there’s almost nothing under the sun” the new service can’t handle.

The cloud-based X8Ms come as single-tenant offerings to avoid noisy or malicious neighbors. They’re the first Exadata systems to break free of “t-shirt sizes” of basically small, medium and large to deliver true resource flexibility—something that has been an emphasis in Oracle Cloud Infrastructure development.

Customers can select almost any configuration of compute, memory and storage, with a minimum configuration of two compute servers and three storage servers to triple-mirror data.

The system elastically scales by adding compute servers at 50 cores a pop, memory in units of 1.3 TB of DRAM, and storage servers each packing 49 TB of database storage and 25 TB of flash.

Loaiza said 86 percent of the Fortune Global 100 rely on Exadata machines, and 28 percent of the Fortune Global 100 use cloud-based Exadata services.

It’s important to distinguish the new offering from Oracle’s Autonomous Database Service, he said.

Autonomous Database offloads all database management to Oracle’s intelligent systems, whereas the X8M service allows customers to stay in the driver’s seat.

“My steering wheel is still there, my brake is still there. Anytime I can take over control,” Loaiza said.