What is a key difference between an on-demand instance and a reserved instance?

On-demand instances are billed by the hour or second (depending on the service) and do not require any long-term commitment, making them a flexible option for users who need computing resources on a temporary basis. This pay-as-you-go model allows businesses to scale capacity up or down according to their needs without the obligation of a contract. Users only pay for the compute capacity they use, which is ideal for workloads with unpredictable traffic.

In contrast, reserved instances are designed for users who can commit to using a specific number of instances for a period, typically one or three years. In exchange for this commitment, users receive a significant discount over the on-demand pricing. This makes reserved instances more cost-effective for predictable workloads. However, this arrangement does not provide the same flexibility as on-demand instances.

Thus, understanding that on-demand instances are billed based on actual usage without long-term commitments is critical for managing costs effectively in cloud environments.