Burst control limits are enforced for subscription tiers in order to distribute the load across the specified time period. For example, if you have a subscription tier that allows you to send 1000 requests per hour, you can ensure that a particular application does not consume the full quota of 1000 requests within the first 2 minutes by setting a burst control limit within the subscription tier allowing only a maximum of 25 requests per minute. Therefore, the time periods set for burst control limits must always be smaller than the time period specified for its corresponding subscription tier. Burst control limits can be set only to control the number of requests for a given period of time and does not allow you to control the data bandwidth for a given time period. The burst control limits are enforced for each individual Gateway node. Although the request counters are replicated across the cluster, since burst control time periods are usually quite small, the replication frequency can be quite high compared to the burst rate of incoming requests. Therefore, it is safe to assume that the burst control values are applied on a per-node basis. For information on how to define burst control limits, see Rate limiting (burst control).
Advanced throttling policies: API Publisher
If advanced throttling is enabled, it will also be applied here. Requests will be allowed/rejected based on the conditions specified in advanced throttling policies as well.