A first request to reserve a quantity of resources that comply with a constraint specified in the first request is obtained. A set of available resources that fulfills the constraint is determined based on current capacity usage. A token is associated with the set of available resources, with the token being associated with an expiration. The quantity of resources from the set of available resources is reserved, where the quantity allocated is unavailable, until the expiration, to fulfill resource requests that lack the token. The token is provided in response to the first request.