Autoscaler, according to an official blog post, is the same system that Google itself uses to dynamically scale the number of VMs being used by a given application based on load – users set utilization targets, and the autoscaling system spins up or shuts down VMs in order to keep, say, RAM utilization at 50%. The idea is to remove the need for extensive capacity planning and management, Google said.
+ ALSO ON NETWORK WORLD: How Comcast is changing tactics in response to cord cutters | How the Red Sox brought new tech to baseball's oldest park +
“This saves you money and headaches since you don’t have to buy and hold spare capacity,” the announcement said. “Furthermore, Autoscaler can scale from zero to millions of requests per second in minutes without the need to pre-warm.”
In addition, VMs on GCE are getting bigger – customers can now use 32-core VMs for load-intensive tasks like video rendering and heavy database workloads. The 32-core machines are available in three flavors – one with lots of RAM, one with comparatively little for more CPU-driven workloads, and another that strikes a balance between the two.
The new features buttress Google’s IaaS play, but the company’s got a long way to go in that particular marketplace, which is dominated by AWS. Gartner’s latest Magic Quadrant report on IaaS showed Amazon with a commanding lead, Microsoft’s Azure in a distant second and Google an arguable third in the trailing pack. Google has been able to get into the market quickly, since it uses much of the same technology to run its own expansive in-house operations, but Gartner said that the company doesn’t yet have the trust of enterprise users.