A diving company requires that divers complete a health questionnaire every 15 days after each dive start. You need autoscaling rules so instances increase while divers fill out the form and decrease afterward. Which two autoscale configurations would satisfy this requirement?

CPU usage-based autoscaling is suitable because an increase in divers filling out forms will directly correlate with increased CPU usage on the application servers. The autoscaling rule can be configured to add instances when CPU usage exceeds a certain threshold and remove them when it drops, effectively handling the fluctuating load. Predictive autoscaling is also a good fit because the requirement states a predictable pattern: "every 15 days after each dive start." This regular, recurring pattern allows predictive autoscaling to anticipate the demand increase and pre-provision resources, ensuring a smooth experience when the forms are due. Recurrence profiles are used for scheduled scaling actions at specific times, not in response to dynamic load or predictable patterns based on events. Fixed date profiles are for one-time scaling events on a specific date, which doesn't fit the recurring nature of the requirement.