Which cost optimization design principle would allow a customer to save on costs by stopping resources from running when they are not in use?

Adopting a consumption model is an effective cost optimization design principle because it allows customers to pay for resources based on actual usage rather than committing to fixed costs. In this model, resources can be scaled up or down according to demand, which means a customer can stop or terminate instances during non-peak hours or when they are not in use. This approach minimizes waste and ensures that the customer only incurs costs for resources actively being utilized.

For example, services such as Amazon EC2 can be configured to automatically stop instances after hours or during scheduled maintenance windows, ensuring that unnecessary costs are avoided. This flexibility in resource management aligns perfectly with the consumption model, resulting in significant savings.

The other options, while they have their benefits, do not directly contribute to stopping resources when not in use in the same way as adopting a consumption model does. Using Reserved Instances involves committing to a specific amount of resources over a longer term, potentially limiting flexibility and not directly addressing the need to stop resources when idle. Leveraging on-premises infrastructure often comes with higher fixed costs and maintenance expenses, which do not inherently allow for scaling down to avoid costs. Purchasing dedicated hosts involves significant upfront costs and ongoing commitments, making it less suitable for cost optimization in variable workloads.