Abstract: A state-space model based on discrete events was proposed to describe the network dynamic scheduling system. A number of heterogeneous jobs were concurrently taken into account. Upon describing the clearing policy of the network dynamic scheduling system, algorithms were developed for analysis of QoS (Quality of Services) performance parameters, which include queue length, total job number, the system's throughput, and response time of job request. Numerical calculations showed that the optimized scheduling policy could make the response time of the system within a controlled scope while the system's throughput was in a stable state.