Last week, Amazon made a huge change in how it charges businesses for its cloud services, saying it would start to bill on a per-second basis starting Oct. 2 instead of by the hour. Now rival Google is also going to per-second increments, but is making the change effective immediately.
In theory, smaller price increments could cut costs for some customers who don’t use a full minute, or full hour, of the computing capacity they have paid for.
Amazon Web Services, Google (googl), and Microsoft have leapfrogged each other on pricing and new cloud computing features for several years now. So this is just the latest chapter in that saga as they fight to get businesses to put more of their data and run more of their software on their respective clouds.
Google’s new price model is for its basic computing units (which it calls virtual machines, or VMs) as well as its container engine and a few other offerings. And the price covers all VMs whether they run Windows Server, Red Hat Linux or SUSE Linux operating systems. Amazon’s per-second pricing applies only to Linux, not to Windows. Google’s pricing on “persistant disk” storage attached to these VMs has been billed per second for quite some time.
Get Data Sheet, Fortune’s daily tech newsletter.
At this week’s Ignite tech conference in Orlando, Microsoft (msft) took another route to price cuts by announcing Amazon-like reserved computing instances that give customers discounts (of up to 72% of of list price) if they lock into using those computing resources for one- or three-year periods.