System and method for routing service requests

Service requests may be routed to services in a service framework provided by a plurality of hosts. A first candidate host may receive a service request. The service latency of the first candidate host may be determined. The service latency for one or more candidate host zones may also be determined. The first candidate host may service the service request or route the service request to one or more candidate host zones. In some embodiments, a probabilistic distribution based at least in part on the service latency of the first candidate host and the service latency for each of the one or more candidate host zones is used to determine whether the first candidate host services the service request or routes the service request to the one or more candidate host zones.