Design And Implementation Of Out-of-place Update Mechanism For Optical Disc

Much attention has been paid to long-term data preservation in a reliable and safe way.With the development of Blu-ray disc technology,the optical disc storage has many advantages such as low energy consumption,low cost,high capacity,high reliability,long-term preservation,ease maintenance,security.The Blu-ray disc library which can automatically manage ten thousands of discs has become a option to store cold data in long-term.But the WORM disc can not realize the in-place update like the traditional hard disks.Thus the implementation of out-of-place update mechanism in optical disc is needed to provide compatibility towards the traditional access pattern,further achieve the provenance of historical data.Aiming at the need for data explicit update and implicit provenance in the optical discs,the pseudo-covering append recording method is designed to achieve logical in-place update semantic and the corresponding procedure of data read.Specifically,the out-of-place update mechanism is implemented on four aspect: firstly,the design of the overall system architecture including write and read module,achieving the transparence in update procedure;secondly,implement the procedure and data structure of multiple-time recording in single disc,the incremental and full append write method is proposed,the file and the update files are written to discs in batches.Whether to do full write or incremental write is automatically chosen based on the file size,thus improving the write performance of the optical discs;thirdly,the logical data update semantic is implemented,designing of the self-defined metadata,recording the append file information in file update,ensuring the consistency of file versions in the optical discs;finally,the file provenance is implemented in the optical discs,the history version of the file can be located.The function test and performance evaluation is done to validate the basic function,and the performance under multiple workload is tested.The result shows that,in the best case,it takes 3.7886 s to read a small file with one time update.It takes 5.3883 s to read a64 MB file with one time update.Compared to the files with no update,the small file access time just increases by 9%,the time of big file just increases by 6%.