For mixed-media workload, disk scheduling strategies have to guarantee the QoS (Quality of Service) of requests with timing constraints while optimizing the disk utilization. In this paper, we present a novel real-time disk scheduling algorithm which optimizes the seek time overhead of all requests while meeting the deadlines of requests with timing constraints. Our algorithm first arranges the requests in the queue by SCAN order and clusters the several adjacent requests into group. And then it finds a feasible schedule which meets the different QoS of each requests using branch-and-bound strategy. Through trace-driven simulation, we show that our algorithm outperforms other algorithms in terms of response time, throughput, and QoS guarantee for real-time requests.