Monitor capacity fulfillment
Capacity fulfillment metrics answer two questions about your Amazon EC2 capacity: how much of your requested capacity Amazon EC2 fulfills, and whether that capacity lands on your preferred instance types and Availability Zones. Capacity Manager organizes the metrics into two groups: fulfillment metrics and the instance type and Availability Zone breakdown. Both are available in the Capacity fulfillment tab of the Amazon EC2 Capacity Manager console and through the get-capacity-manager-metric-data API. Together, they help you analyze your capacity flexibility, so you can see where it is paying off and where to add more.
Note
Capacity fulfillment metrics are available only to AWS Support customers on the Unified Operations plan.
Start at the organization level for a summary, then use dimensions such as tag, account, or Region to drill into the specific workloads where you need to act. Capacity fulfillment metrics are available through get-capacity-manager-metric-data with a minimum period of one hour.
How to read the two groups together
Capacity is measured at two levels, and you read them in order. Request-level fulfillment (the top level) answers the first question: how much of the requested capacity does Amazon EC2 fulfill? A request can be flexible across multiple instance type and Availability Zone combinations, where a combination is a single instance type in a single Availability Zone. A flexible request is fulfilled if any of those combinations delivers. For example, suppose you have a request that is flexible across ten combinations. It doesn't matter whether Amazon EC2 fulfills the capacity on the first instance type or the tenth. The request still counts as 100% fulfilled. A high request-level fulfillment rate is exactly what you want to see: it means your flexibility is working.
Per-combination fulfillment (the second level) takes each option within a flexible request into account, measuring how each individual instance type and Availability Zone combination performed. These rates are often lower than the request-level rate, because they don't include the benefit of a fallback (an alternate instance type and Availability Zone combination that Amazon EC2 uses when your preferred option isn't available).
Start with the request-level metrics to confirm that your requested capacity is being fulfilled. If it is, your flexibility is doing its job. Then use the per-combination metrics for optimization — to see where your capacity is landing and decide whether to reserve capacity or adjust your instance type choices.
