In existing multi-job admission control schemes with waiting queues for reducing the response time of soft real-time jobs, the service failure rate and response time distribution that explicitly account for allowable delay have not been sufficiently analyzed. However, for applications requiring high reliability and low latency, the lack of explicit QoS guarantees can significantly degrade service quality. This paper proposes an admission control scheme and an efficient control-parameter search method that analytically derive job response times, thereby minimizing the expected response time of soft real-time jobs while satisfying a service failure rate constraint defined based on the job rejection rate and the delay violation rate conditioned on successful admission. Numerical results demonstrate that the proposed method enables efficient parameter configuration without performance degradation.