Embodiments of the present invention address deficiencies of the art in respect to workload management of backend systems and provide a novel and non-obvious method, system and computer program product for limiting a request rate from a client based on a health state of a server system. In one embodiment of the invention, a computer-implemented method for limiting a request rate from a client based on a health state of a server system that includes monitoring a server system, determining a measure of health of the server system, sending the determined measure of health to a client submitting a request to the server system and receiving a subsequent request from the client that is delayed by a time period, wherein the time period is based at least in part upon the determined measure of health. The computer-implemented method can further include receiving server system CPU utilization, determining an approximate predicted real server response time of a transaction and sending the determined approximate predicted real server response time of a transaction to a client submitting a request to the server system.