作者: William V. Courtright Ii , Mark Holland , Jim Zelenka , Garth A. Gibson
关键词: RAID 、 Computer hardware 、 Unix 、 Embedded system 、 Computer data storage 、 Code (cryptography) 、 File system 、 Computer science 、 Disk array 、 Extensibility 、 Scheme (programming language)
摘要: Error recovery in redundant disk arrays is typically performed an ad hoc fashion, requiring architecture-specific code which limits extensibility and difficult to verify. In this paper, we describe a technique for automating the execution of array operations, including from errors, independent architecture. Our approach employs graphical representation operations two-phase error-recovery scheme refer as roll-away error recovery. We demonstrate validity RAIDframe, prototyping framework that separates architectural policy mechanism. RAIDframe facilitates rapid proto- typing new RAID architectures by localizing modifications. addition, RAIDframe-implemented run same when configured event-driven simulator, user-level application managing raw disks, Digital Unix device-driver capable mounting file system. Evaluation shows performance equivalent less complex implementations case studies levels 0, 1, 4, 5, 6, parity declustering achieve expected