Google’s Container Engine, its service for running and managing Docker containers on its cloud platform using its open source Kubernetes system, is now out of beta and generally available. This means Google now considers it ready for production use and backs it with a 99.5 percent uptime SLA.
Google made an early bet on containers for running its own data centers and over the last year or so, it started making some of what it learned from running containers at this massive scale (the company says it launches more than 2 billion container instances across its data centers every week) available to others. The prime example for this is Kubernetes, its container management tool, which it recently donated to the newly formed Cloud Native Computing Foundation.
Container Engine builds on this work and allows developers to set up a managed cluster for their container deployments with just a few clicks. Google notes that as companies like Red Hat, Microsoft, IBM, Mirantis and VMWare start integrating Kubernetes into their platforms (often as part of their OpenStack platforms), developers will be able to easily port their workloads between cloud providers as needed.
Developers who want to use the service need to set up their clusters and declare their container’s requirements (including how much CPU and memory they will need, for example). From then on, Google’s service will follow these instructions and monitor the cluster. Google offers its own Container Registry, which reached general availability two months ago, for storing and accessing private Docker images and developers can scale their cluster as needed. The system also allows for hybrid deployments using the Google Cloud VPN.
Using the service is free for basic clusters with up to 5 virtual machines (though you still include the cost of running your Compute Engine instances!). Standard clusters with up to 100 virtual machines costs $0.15 per hour (plus the Compute Engine and other Google Cloud Platform cost).
Correction: Google previously said the price per hour for running up to 100 machines was $0.015 per hour. It’s $0.15. The company also previous said the SLA was for a 99.95 percent uptime. It’s 99.5 percent. We’ve updated the post to reflect this.