Currently, APIMan seems to expose an overall limit to the number of requests across an entire API for both Quota and Rate Limit policies. This may not be practical for commerce APIs.

Given a shopping API that has multiple resources:

/search
/availability
/book

It is very likely that we will want to provide varying levels of limits / quotas per resource.

Imagine all our quota/throttle limits are consumed by /search requests, then the /book endpoint will not be available to the API Client (due to the quota being exceeded) so a client will be prevented from making a booking via the /book resource. This will result in lost revenue for an API provider due to overall throttling.

It is better to provide per URI / resource level throttling so that various quotas can be assigned to a more granular level.

Hope this clarifies.