vCPU and OCPU pricing information

August 25, 2021 | 4 minute read
Dan Reger
Senior Manager, Product Marketing, OCI
Text Size 100%:

To help customers more easily compare competing products and better align to this common industry practice, Oracle is changing the way we express pricing on the web for compute-based products to the industry-standard vCPU approach. Our underlying price list and billing remain aligned to OCPUs. So why the change?

Clarity for customers

The model other clouds use to measure their compute resources is virtual CPU or vCPU, and it’s truly virtual. An end user has no idea what resources underlie that abstraction, or how much a cloud provider might have oversubscribed their physical network bandwidth and compute cores for efficiency—theirs, not yours. Because of this variability, the noisy neighbor problem, and other factors, these clouds provide you with no guarantee of performance.

When we designed Oracle Cloud Infrastructure (OCI), our goal of ensuring predictable, high performance resulted in design principles of non-blocking networks and non-oversubscribed compute. As a unit of measurement, vCPUs didn’t make sense for our products.

OCI customers receive a real fraction of the underlying physical infrastructure. For example, one of our VM.Standard2.2 Compute instances provides two OCPUs, which represent two of the total 52 physical cores in the system, which supports the simultaneous run of four threads. It also includes 30 of the 768 GB of RAM and 2 Gbps of the 50 Gbps available to the underlying system. Not much is virtual about that approach. You get a real piece of the underlying hardware, and we back that with performance and manageability SLAs. Most clouds don’t.

How OCPU is better for you

The Oracle CPU (OCPU) unit of measurement for x86 OCPUs is worth at least two vCPUs. That’s only counting two cores per physical CPU to account for two threads of execution: The main CPU core and its associated symmetric multiprocessing (SMP) unit. Think of SMP as a tiny core within a core that can run a second task (thread) in parallel while sharing some resources with the main CPU core.

In reality, most cloud compute substantially oversubscribes both compute and networking resources and depends on customers’ underutilization to keep things up and running. If most or all the virtual machines (VMs) on a particular physical server in most clouds used the full resources they were granted, the system would grind ungracefully to a halt as processes waited longer and longer to run and network requests backed up.

At Oracle, we call customers VMs fully utilizing their system resources "Just another Monday." We expect our customers to use the resources that they purchase for the workloads they run on our cloud. No company wants business-critical applications like Oracle Database or E-Business Suite to stop working under a load because their public cloud provider has oversubscribed the underlying infrastructure. All businesses work to ensure that their workloads have enough resources when running on-premises, and Oracle wants to ensure that you can depend on that fact in the cloud.

So, now you understand OCPUs, what they mean, and why we use them in our products and our pricing. So, why the change in the way we express pricing?

Oracle Compute costs roughly half as much as AWS for several years now. But because our OCPU represents at least two vCPUs (or more), it’s hard for people to see that quickly and easily without first learning about our architecture. The following table shows a direct comparison using differing terminology:

A table showing the comparison in price, processors, memory, and network bandwidth between Oracle VM.Standard.E3.Flex and AWS.

While you’re clearly paying less with Oracle, it’s less clear, especially to those who aren’t familiar with OCPU, that you’re getting roughly the same or more compute in the first Oracle row as you are with the AWS option and double the AWS compute with the second Oracle option. Recasting the same data in industry standard terms makes Oracle’s advantage clear.

A table comparing Oracle VM.Standard.E3.Flex and AWS for vCPU.

With Oracle you save over 50%, have performance and manageability SLAs that AWS doesn’t provide, and are running your workload on a cloud that expects, not fears, that you can fully utilize the resources you’ve purchased. And now it’s easier to make these direct comparisons.

Pricing, and the official price list document, remains unchanged. Contracts are unchanged. The product is unchanged. Metering is unchanged. Bills are unchanged. But now Oracle Cloud Infrastructure web pages display pricing in terms of OCPU and vCPU to make price comparisons easier.

To learn more about our pricing, see our Pricing page.

Dan Reger

Senior Manager, Product Marketing, OCI

Previous Post

Monitor cloud security proactively with email alerts

Rachel Lam | 3 min read

Next Post

Five requirements to consider when evaluating cloud providers for GxP workloads

Andrew Hahn | 4 min read