We research a broad spectrum of issues to enhance the performance of I/O and the reliability of data in single or distributed systems. By bridging the gap between OS storage stack and storage device, we make the best use of the new features of recent storage devices. Intelligent resource-scheduling enables a distributed system to process data with high resource utilization. In addition, our research focuses on applying protection mechanisms to various software layers to prevent critical data loss. The direction of our work will lead to high-performance/high-reliability computer systems and satisfy the diverse requirements of users.