Oracle added capabilities to its public cloud geared toward enterprises running infrastructure-heavy workloads including high-performance computing (HPC), big data, and artificial intelligence (AI).

The company has been adding performance and storage features to its public cloud – Oracle Cloud Infrastructure – as it tries to unseat market heavyweight Amazon Web Services (AWS). Oracle aims to make its cloud “the only cloud you’ll ever need to run your entire organization,” said Dave Donatelli, executive vice president worldwide sales and marketing strategy at Oracle Cloud solutions, at last month’s Oracle OpenWorld in San Francisco.

The new compute instances —  virtual machine (VM) and bare metal compute, and new bare metal graphical processing unit (GPU) instances — are based on the company’s X7 hardware. Oracle claims the upgrades best AWS with 1,214 percent better storage performance at 88 percent lower cost per input/output operation.

“One of the fundamental things we’ve done is we’ve made a bet on storage,” said Leo Leung, director of product and strategy at Oracle Cloud Infrastructure. This bet on storage is what allows the company to make its higher performance and lower cost claims, he said.

Both VM and bare metal instances support non-volatile memory express (NVMe) solid state drive storage. NVMe is super fast flash technology.

“We think storage is strategic and we’re going to make investments to make our storage the fastest available in the market,” Leung said. “Some people talk about processor power, they talk about teraflops at the compute level, but ultimately you are going to have to write these things into storage. If your storage is slow, it doesn’t matter how fast your compute is because that will be the bottleneck.”

The cloud advancements also mean customers don’t have to buy additional on premises hardware for compute-heavy workloads, Leung added.

“If you look at HPC of AI types of projects, they are trying to test something or run a set of data through a model and there’s a very specific endpoint," Leung said. "The ability to just spin these things up in a minute, do the job, and then turn it off is a huge business advantage. They just run the project and pay for what they use.”

The new compute instances are powered by Intel’s latest Xeon processors based on the Skylake architecture. The accelerated bare metal shapes are also powered by NVIDIA Tesla P100 GPUs, based on the Pascal architecture.