From the Oracle Blog:
We’re happy to announce that today we’ve enabled per-second billing for all Compute instances and Autonomous Database usage on Oracle Cloud Infrastructure. From its inception, Oracle Cloud Infrastructure was built to meet the needs of demanding enterprises in running their most important workloads. With this announcement, we’re making Oracle Cloud Infrastructure a stronger platform for legacy applications moving to cloud and a new family of cloud native applications that rely on microservices and dynamic scaling.
With this billing model, usage of Compute and Autonomous Database is billed per-second. All prices continue to be quoted on an hourly basis.
Here are some details about this billing model:
- All virtual machine (VM) Compute instances, including those with graphical processing unit (GPU) chips, are now billed per-second with a one-minute minimum.
- All bare metal Compute instances, including those with GPU chips, are now billed per-second with a one-hour minimum.
- Autonomous Data Warehouse and Autonomous Transaction Processing usage is now billed per-second with a one-minute minimum.
- Windows OS images are now billed per-second with a one-minute minimum.
- Microsoft SQL Server images available in the Oracle Cloud Marketplace are now billed per-second with a 744-hour (one month) minimum.
Use Cases with the Biggest Expected Impact
The workloads that will see the biggest impact from this change are those that stop and start frequently, and those that run for short durations. Here are some examples where we think customers will see significant benefits:
- Application and database development: When cloud applications are being developed, it’s common to create and destroy resources repeatedly. For example, a developer who’s working on a complex deployment task might run parts of a deployment script that deploys an instance and database 10 times in an hour. Previously, each time an instance was launched, it was billed for a minimum of one hour. Now, customers pay for their actual usage of fractions of hours and minutes.
- High-performance computing (HPC): Modeling and rendering jobs require a large amount of computing resources for a relatively short time. HPC orchestration tools give customers the choice of running intensive jobs over a large number of instances for a short time or over a smaller number of instances for a longer time. Without per-second billing, rounding could lead to increased charges that can add up across many jobs.
- Continuous integration/continuous development (CI/CD): Integration and development pipelines use temporary instances as worker nodes to perform small, discrete tasks. For example, a workload might create a temporary VM to compile a build or run a deployment task that gets deleted when that task is complete. Because these temporary resources are often used for less than an hour and across a portfolio of applications, the benefit from per-second billing can be significant.
See It for Yourself
Per-second billing is live now. You can expect to see the impact of the change reflected automatically on your next bill—no action is required on your part. Use Cost Analysis to see the high-level impact and Usage Reports to see resource-level data.
Image: Fractional Hour Usage in Cost Analysis