Amazon Web Services To Bill By The Second For Virtual Linux Servers

Amazon Web Services upped the ante in the cloud price wars Monday by saying it would become the first major public cloud operator to start billing customers by the second for virtual servers and storage.

AWS will measure usage down to the second for EC2 compute instances and EBS (Elastic Block Store) volumes. The public cloud leader currently offers per-hour metering, while the next two trailing competitors, Google Cloud Platform and Microsoft Azure, have for several years charged by the minute.

Amazon's new billing model is only available for Linux machines and does have a 1-minute minimum charge. It applies to On-Demand, Reserved and Spot instances, according to Jeff Barr, AWS' chief evangelist.

[Related: AWS Warns Users To Secure Storage Buckets]

"Many of our customers are dreaming up applications for EC2 that can make good use of a large number of instances for shorter amounts of time, sometimes just a few minutes," Barr wrote on the AWS blog.

Workloads taking advantage of AWS EMR, a service offering managed big data frameworks on scalable EC2 clusters; and AWS Batch, which enables large batch processing jobs, also qualify to be billed by the second.

Per-second billing will be available in all AWS regions on October 2. AWS prices will still be listed by the hour, but customers will see bills in which their duration of usage contains decimal points, Barr said.

Jeff Aden, executive vice president of marketing and strategic development at 2nd Watch, an AWS partner based in Seattle, told CRN the new model will not only allow some customers to save more on their cloud spending, but will also create opportunities for partners to help them do it.

Per-second billing will primarily reduce costs for customers running large, complex environments. In many cases, however, to realize those savings will take another layer of management, provided either through automation tooling or partners, Aden told CRN.

"This is a big move on [AWS'] part," Aden said. "It becomes more valuable the more complex the environment becomes."

Cost-saving methodologies employed by enterprises, like auto-scaling or instance parking, can offer more value with more-granular billing, Aden said.

When an environment automatically scales up, creating new instances to deal with short usage spikes, the current system bills each new server for an hour, even if it only existed for minutes, Aden said.

And there will be more potential to reduce AWS spending from parking—turning off servers while not in use.

"The more you can micro-tune it, the more savings you can get. If you have many thousands of instances running, you could have a material impact," Aden told CRN.

It's important to remember the new pricing model only applies to Linux-powered virtual servers. While most tech companies deploy Linux on AWS, 2nd Watch sees a pretty even split among enterprises between Windows and Linux, Aden said.

Barr noted that while per-second billing may amount to a price reduction for many workloads, a more important aspect of the change will be how it motivates users to think differently about approaching projects that require high computing capacity, such as 3-D rendering.

With the new model, the cloud leader leapfrogs its main competitors when it comes to billing granularity.

Google, then shortly after Microsoft, introduced per-minute billing back in 2013. But despite the one-upmanship of the cloud wars, AWS never followed suit and maintained the hourly billing model.

Earlier this summer, Microsoft introduced a simplified container service called Azure Container Instances, the only service on Azure so far to bill by the second.

The way cloud providers bill their customers is being impacted by new architecture paradigms, especially serverless computing services, such as Amazon Lambda, that simply charge customers for operations executed, Barr noted.