Emerging video surveillance, large-scale sensor networks, andstorage-bound Web applications require large, high-performance, andreliable storage systems with high data-throughput as well as shortresponse times for interactive requests. These conflictingrequirements call for quality of service (QoS) support. These storagesystems are often implemented using Redundant Arrays of IndependentDisks (RAID). In this paper we investigate the effectiveness ofpreemptive disk-scheduling algorithms to achieve betterQoS. We present an architecture for QoS-aware RAID systems based onSemi-preemptible IO. We show when and to preempt IOs to improve the overall QoS of the RAID. Using our simulator for preemptible RAID systems, we evaluate the benefits and estimate the overhead of the proposed approach.