In this paper, we present the design and implementation details of the Open ROW Snapshot which is the data backup scheme based on the snapshot approach. As the data to be stored in storage systems are tremendously increased, data protection techniques have become more important to provide data availability and reliability. Snapshot is one of such data protection techniques and has been adopted to many file systems. However, in large-scale storage systems, adopting a snapshot technique to prevent data loss from intentional/accidental intrusion is not an easy task because the data size being backup-ed at a given time interval may be huge. In this paper, we present the Open ROW Snapshot that has been implemented based on the file system-based snapshot approach. The Open ROW Snapshot provides a widely portable structure and causes less I/O processing overhead than the ROW (Redirect-On Write) method does. Furthermore, the Open ROW Snapshot provides a capability of maintaining the disk space assigned to snapshot images in a consistently-sized disk portion. We present the performance results of the Open ROW Snapshot obtained from the Linux cluster located at Sejong University.
展开▼