Dirty region tracking nonvolatile media

Storage set (e.g. an array of hard disk drive), writing to the memory set is in progress, the power loss may experience a failure such as disconnection of a software crash or storage device. Recovery from failure to detect inconsistency may include scanning the stored set to correct (e.g., inspecting or checksum comparing mirror each other data set). However, the lack of information about the position of the write pending to the memory set during failure, the "cleaning" process will be scanning the entire memory set, the recovery process that is stretched as a result It could become. Presented in this paper is distributed to various areas of the region size with the stored set (for example one gigabyte), storing the descriptors of "dirty" region, including a write in progress on a non-volatile storage medium it is a technique that tracks the writing to the memory set by. That way, the recovery process after a failure may be limited to regions identified as dirty.