Oracle’s Cloud Roadmap Features New High-Power Instances

Looking to jump into the IaaS leadership pack, Oracle next year will deliver instances powered by cutting-edge chips from Intel, Nvidia, Ampere and AMD that offer unique flexibility in resource allotment.

Oracle unveiled its coming lineup of cloud infrastructure services Tuesday, showcasing four new instances at the center of a cloud strategy focused on powering compute-intensive workloads with unrivaled flexibility.

Oracle will outfit its public infrastructure cloud, which has gone in six years from skunkworks project to spanning two dozen regions globally, with cutting-edge processors from Intel, AMD, Nvidia and Ampere, Karan Batta, vice president of product management for Oracle Cloud Infrastructure, told CRN before the event.

Those hardware options, most slated to come online early next year, will improve the cost-performance dynamics for customers running high performance computing (HPC), artificial intelligence, edge and traditional workloads on Oracle Cloud Infrastructure, Batta said.

[Related: Xactly Chooses Oracle Cloud To Host Its SaaS]

Oracle Cloud Chief Clay Magouyrk introduced the new instances in an online keynote joined by Intel CEO Bob Swan, Nvidia CEO Jensen Huang, and Ampere CEO Renee James.

While Oracle isn’t the only provider to deliver the latest chipsets as cloud services, its cloud enables customers to access them with unique flexibility, Batta said.

“Processors themselves are not so much a differentiator as the platform itself,” he told CRN.

OCI is the only cloud that allows right-sizing environments by granularly selecting the number of cores and memory for optimal economics.

“We’re moving to an iteration where you pick your architecture, pick your cores, pick your memory and you’re good to go. Slice and dice the VM you want rather than be stuck with one that you’re given,” Batta said. “You can’t do that with any other provider.”

OCI also stands out in its ability to support applications built before the days of cloud-native design methodologies, he said.

“Most of our competitors say lift-and-shift,” Batta said, but they demand enterprises refactor applications, or blow them up entirely, if they want to move them to cloud.

Oracle’s mantra of “move-and-improve” highlights that an enterprise can migrate any workload to OCI, regardless of its operating system, network architecture or dependencies.

“You can literally take what you have and bring it over. You have a ten-year-old OS, RDMA for networking, you can move it, then modernize components of your architecture once you’ve moved in.”

Oracle believes a significant migration of HPC workloads to the cloud has begun based on what it’s seeing from large manufacturing and automotive customers like Nissan Motors, which is now running crash simulations and 3D visualizations on tens of thousands of production cores in the OCI Japan region.

To accelerate those migrations, in early 2021 Oracle will replace its X7 instances with new X9s powered by Intel Ice Lake processors. Those latest HPC instances should boost performance by more than 30 percent from the previous generation at roughly the same cost, Batta said.

For customers looking to cost-effectively run general purpose workloads like web applications or microservices deployments, Oracle will introduce new E4 instances, powered by AMD’s Milan generation of processors, early next year. Those build off the E3 generation of AMD Rome processors, which were the first to allow customers to mix-and-match memory and CPU.

Another new offering will be the first ARM instances in OCI, with chips provided by Ampere Computing.

Those also allow variable numbers of cores and memory, with the benefits in cost and power consumption that make ARM architectures well-suited for enterprises investing in mobile and Internet of Things technologies.

Oracle also revealed that on September 30 it will make generally available its next-generation GPU platform powered by Nvidia A100s. The design of those accelerated compute instances was based on feedback from customers increasingly adopting AI in their businesses, Batta told CRN.

Oracle’s GPU hardware differentiates itself in its ability to network large clusters and run them on bare-metal. Oracle also is adding the ability to attach up to 25 terabytes of local storage to those instances.

Oracle was a late entrant to the IaaS market, but the company believes OCI is finally gaining significant momentum because of its cost-dynamics and performance, as well as an optimized ability to run Oracle databases and applications.

While Oracle’s market share still trails Amazon Web Services, Microsoft Azure and other leading providers by a large margin, the field is wide open, Batta said, as most large enterprises are only now starting to think about fully migrating their mission-critical workloads.

“The core systems that provide their business value have not moved to the cloud,” Batta told CRN.

While enterprises want the flexibility of pay-per-use, the ability to scale up and down based on demand, and to avoid the capex risk in adopting new technologies, they’ve been hesitant for good reasons.

Still holding back migrations is the desire for on-premises performance and the ability to fine-tune environments to exact specifications. In short, enterprises want “all the benefits of on-prem without any of the downsides,” Batta said.

For that reason, “we’ve been very methodical in how we invest our time and energy on those things,” Batta told CRN.

That approach to development, demonstrated by the four new OCI instance types, “should only give channel partners more confidence about Oracle’s stance in the cloud,” he said.