| 摘要 |
Special fields for error recovery are provided in data set information tables of data sets written with synchronized transactions. If a transaction only partially fills a data set, that data set is rewritten in a succeeding data set, appending the next transaction. A moving access point in the table identifies the appended transaction, allowing the rewritten transaction to be skipped during read recovery. The table provides recovery trails by providing a thread to tie the data sets together, the status of the data set, and pointers, such as identifying the wrap of the immediately succeeding data set.
|
