In an exciting, new announcement on their blog, Amazon Web Services (AWS) has just announced that EC2 Instances running Linux, and “are launched in On-Demand, Reserved, and Spot form,” will support per-second billing. This includes AWS Batch and Elastic Map Reduce (EMR). Since 2006, when EC2 was launched, EC2 Instances were elastically billed by the hour. Now, that has all changed, and AWS customers are given even more granularity with their purchasing options.
In addition, provisioned storage for Elastic Block Store (EBS) volumes will benefit from one-second incremental billing.
As you can imagine, when deploying workloads onto EC2 instances that run for less than a one-hour period, you will be capable of realizing a significant cost savings using per-second billing. There are other services in AWS that support fractional billing as well, although they are unaffected by the announcement about per-second billing for EC2 Instances. For example, AWS Lambda enables customers to deploy code without having to concern themselves with deploying, managing, or monitoring the operating system or virtual hardware of the underlying instance. As noted in the pricing page for AWS Lambda, billing for Lambda function execution is handled at per-100 millisecond intervals. Customers running workloads in Amazon EC2 Container Service (ECS) will potentially see cost savings for container instances that are running for fractional hours.
As you explore the new or continued use of Amazon EC2, keep the following key points in mind:
- Per-second billing will apply to pre-existing Linux EC2 Instances and new ones.
- Per-second billing is effective on October 2nd, 2017.
- Prices for EC2 Spot Instances are calculated hourly, but billed fractionally at second-level precision.
- EC2 instances running Microsoft Windows Server currently do not support per-second billing.
- From the linked article: “The Dedicated Per Region Fee, EBS Snapshots, and products in AWS Marketplace are still billed on an hourly basis.“