Start/Stop Automation: Pay Only For What You Use

Why it exists

Cloud providers bill for running resources, even if they're idle. Many environments, especially for development and testing, are only used for 8-10 hours a day, 5 days a week. Leaving them on 24/7 means paying for over 100 hours of idle time per week, which can account for over 70% of the bill for those resources.

The mental model

Think of it as an automatic light switch for your cloud infrastructure. You wouldn't leave every light in an office building on 24/7. You put them on timers or motion sensors. Start/stop automation does the same for your virtual machines and databases, turning them off when no one is "in the office" to save on your cloud bill.

How it works

You tag your cloud resources (e.g., VMs, databases) with a specific schedule name, like "uk-office-hours". A central scheduling service, like AWS Instance Scheduler or Azure Automation, runs on a timer. It scans for resources with schedule tags. If the current time falls within the "stop" period for a resource's schedule, the service issues a 'stop' API call. When the "start" period begins, it issues a 'start' call. Schedules are often defined using cron-like expressions and must handle timezones carefully.

When to use it

This is a must-have for non-production environments like development, testing, and staging. It's also useful for batch processing workloads that only need to run for a few hours a day, or for demo environments that are only spun up for specific presentations.

When not to use it

Do not apply this to production services that require 24/7 uptime. Also, avoid it for stateful applications that cannot tolerate being shut down or that take a very long time to restart. For example, a database with a massive in-memory cache might perform poorly for a long time after each startup, negating the cost savings with poor performance.

One canonical example

A development team has 20 EC2 instances for their work. Using AWS Instance Scheduler, they tag these instances with Schedule=uk-office-hours. They define this schedule to run from 9:00 AM to 6:00 PM, Monday to Friday, in the Europe/London timezone. The scheduler automatically stops the instances at 6:01 PM and starts them at 8:59 AM, reducing their EC2 costs for these instances by roughly 70%.