Resource Scheduling

Resource scheduling turns cloud resources off when nobody needs them. Many environments, especially development, testing, training and demo systems, are only used during working hours. Left running around the clock, they cost money for nights and weekends when no one is using them. Scheduling sets these resources to stop at the end of the working day and start again before people arrive, cutting their running hours, and therefore their compute costs, substantially.

Scheduling is one of the simplest and most reliable usage optimization techniques. A typical working week of around 50 hours is less than a third of the 168 hours in a week, so scheduling non-production resources can reduce their compute costs by well over half. For IT financial management, scheduling is a low-risk way to cut waste that can be automated and enforced with policies, and it is often one of the first optimizations a FinOps team introduces.

Definition of Resource Scheduling

Resource scheduling is the automated starting, stopping or scaling down of cloud resources according to a predefined timetable that reflects when they are needed. It applies mainly to resources billed by running time, such as virtual machines, container clusters and some databases.

It is part of usage optimization in cloud cost optimization and addresses one of the most common sources of cloud waste: non-production resources running when nobody is using them.

What Can Be Scheduled

Development and test environments are the most common candidates, since they are typically used only by engineers during working hours. Training, demo and sandbox environments are similar.

Some production workloads can also be scheduled, such as internal tools used only during business hours or batch processing systems that run at fixed times. These need closer agreement with their users before schedules are applied.

Resources billed by storage rather than running time, such as storage volumes and object storage, continue to incur charges when compute is stopped, so scheduling reduces but does not eliminate the cost of an environment.

How Scheduling Works

Schedules are usually defined by tags, such as a schedule tag specifying office hours in a given time zone, which automation tools read to start and stop resources.

Cloud providers offer native scheduling tools, and many FinOps and automation platforms provide scheduling with features such as calendars, holiday rules and self-service overrides. Holiday rules matter, since environments often run unused over public holidays.

For container platforms, scheduling can scale clusters or workloads down to zero outside working hours and back up in the morning.

Importance of Resource Scheduling in IT Financial Management

Non-production environments can represent a significant share of cloud spending, especially in organizations with many development teams. Running them continuously wastes much of that spending.

Scheduling recovers it with little risk. The savings are predictable, the effort is modest and the change is easy to reverse if a schedule turns out to be wrong.

Because non-production resources do not serve customers, stopping them outside working hours rarely affects the business, provided teams can override schedules when they need to work late.

Predictable Savings

The savings from scheduling are easy to estimate. Reducing running hours from 168 to around 50 a week cuts compute costs for scheduled resources by about 70%, before accounting for storage and other always-on components.

Savings continue automatically once schedules are in place, unlike one-time cleanups that need to be repeated.

Scheduling also reduces the base of usage that might otherwise be covered by commitments, so it should be applied before buying reserved instances or savings plans for non-production workloads.

Cultural Benefits

Scheduling makes cost visible in a concrete way. Engineers who see their environments stop in the evening become more aware of what running resources costs.

Providing an easy override, such as a self-service button or tag change, respects engineers' need to work flexibly while keeping the default efficient.

Scheduling depends on reliable environment and owner tags, so it reinforces a good tagging strategy.