| 摘要 |
A primary server transmits a write transaction to N backup devices. Each backup device records the write transaction to a local redo log stored at a high-performance storage and then transmits a notification to the primary device. When the primary device receives notifications from M of the N backup devices, it commits the write transaction, recording it to a low-performance storage. Once the write transaction has been recorded to the low-performance storage, the primary device requests that the backup devices delete their logs. If the primary device incurs a fault, one of the backup devices becomes the new primary device and determines which backup device has the newest local redo log. The backup device having the newest log sends it to the other backup devices, which update their local redo logs to the same state as the newest log. Thereafter, the new primary device is ready to resume transaction processing.
|
