Practical guide
EC2 business-hours scheduling: calculate the compute savings
estimate EC2 savings from stopping development instances. Includes a worked example, calculator scope, and practical questions.
Updated · Sources checked
How much can a schedule save?
For an On-Demand development instance, the compute reduction follows the running hours you remove. A schedule saves money only when the instance can actually stop during the excluded period. It does not reduce an existing commitment simply because an application is idle.
Start with a workload that has a predictable workday: a test server, preview environment, or internal development box. Record who needs it outside office hours and whether overnight jobs depend on it. A calendar assumption is not a utilization measurement.
Worked example: three development servers
Assume three identical instances run for 730 hours in the baseline month. They consume 2,190 instance-hours. A proposed schedule runs each instance for 10 hours on 22 working days: 220 hours each, or 660 instance-hours in total.
The difference is 1,530 instance-hours. Dividing that by the 2,190-hour baseline gives approximately 69.9% less compute usage. Multiply 1,530 by the matching On-Demand hourly rate to estimate the compute savings. This percentage describes the example schedule, not a guaranteed reduction in the AWS bill.
Enter three instances and 730 hours in the calculator, then repeat with 220 hours. Use the same instance type in both runs so the comparison isolates scheduling. A longer month, regional holiday, or emergency weekend changes the result.
Expenses that survive a stop
Attached EBS storage can remain billable while compute is stopped. Public IPv4 resources, backups, load balancers, monitoring, and other services also need separate review. The calculator models Linux t3 On-Demand compute in US East; it does not estimate those attached services or CPU-credit charges.
Before adopting the schedule, test restart behavior, application readiness, and access to persisted data. Keep recovery time in the decision: a server that saves compute but repeatedly delays a team may not be an improvement.
Frequently asked questions
Is 730 hours the length of every month? No. It is a planning convention. Use the actual running hours for a calendar-specific forecast.
Does stopping an instance cancel a Reserved Instance or Savings Plan? No. Evaluate whether other eligible usage consumes that commitment before counting cash savings.
Calculate your scenario
Use the ec2 cost calculator. Keep the displayed region, pricing date, billing units, and exclusions alongside your result. The arithmetic examples above illustrate usage or explicitly hypothetical rates; they are not AWS quotes.
Source
Billing structure checked against official AWS documentation on September 6, 2026.