Skip to content

Q9. Abide no idle compute resources

Billable charges accrue for a compute resource whenever it is in a running state. Just as we do not wish to pay for electricity to provide lighting in an unoccupied building, we want to avoid idle (or underutilized) resources as much as possible. Idle resources have a cost, yet deliver no business value.

The most common compute resources are virtual machine instances. However, platform services that require running compute resources, such as a full-text search engine service, are also a potential trouble spot for idle resources.

There are several strategies that can be used to avoid incurring charges for idle resources.

  • Auto-scaling mechanisms can be used to scale computing resources according to demand.
  • Computing resources used only during business hours could be shut down outside of business hours using a platform's "scheduled task" functionality. Candidate resources might include pre-production, staging, or development instances.
  • Platform services can be used as a substitute for infrastructure resources such as virtual machines. For example, running containers using the provider's orchestration service and compute pool may relieve your team of having to predict and manage the scale of the compute capacity needed.