Adaptive high-performance database redo log synchronization

摘要

A method, system, and computer program product for adaptive high-performance database redo log synchronization. The method commences upon performing a write operation of a redo log entry, the write operation concluding with an indication of completion of the write operation of the redo log entry. Any number of committing processes may be waiting for the indication of completion, and upon indication of completion, then (using a first synchronization mode) the processes or proxy measures the waiting time as experienced by the committing processes (e.g., while waiting for the indication of completion of the write operation of the redo log entry). In some cases a second synchronization mode would introduce less latency than the first synchronization mode, so the system changes to a second synchronization mode. The system can also change mode when a predicted second mode waiting time is smaller than the measured waiting time.

主权项

1. A computer implemented method for adaptive high-performance database redo log synchronization, the method comprising:
performing a write-ahead logging operation comprising a plurality of committing processes that store a number of redo log entries; establishing a post-poll mode indication value comprising at least a poll-wait sense and a post-wait sense wherein the poll-wait sense corresponds to a poll-wait mode, and the post-wait sense corresponds to post-wait mode; setting the post-poll indication value to the post-wait sense corresponding to a first synchronization mode in which the write-ahead logging operation is performed; in a first synchronization mode:
determining a first measured waiting time taken by at least one committing process of the plurality of committing processes while waiting for a receipt of an indication of completion of the write-ahead logging operation that stores a redo log entry of the number of redo log entries in a persistent storage and storing the first measured waiting time in a first location of computer memory;computing and storing, at a latency computation module stored in the computer memory, an expected second mode waiting time that a second synchronization mode is predicted to take until the receipt of the indication of completion of the write-ahead logging operation that stores the redo log entry, wherein the expected second mode waiting time is computed based at least in part on an expected number of polls and an expected poll interval time;selecting between the first synchronization mode and the second synchronization mode based in part or in whole upon whether the expected second mode waiting time is smaller than the first measured waiting time, wherein the second synchronization mode is selected when the expected second mode waiting time is smaller than the first measured waiting time; setting the post-poll mode indication value to poll-wait sense corresponding to the second synchronization mode when the second synchronization mode is selected; and implementing a durability property providing no loss of data from failures in storing the number of redo log entries in the persistent storage at least by committing the write-ahead logging operation after waiting for and receiving the receipt of the indication for successful completion of storing the number of redo log entries in the persistent storage, regardless of whether the first synchronization mode or the second synchronization mode is selected to service the write-ahead logging operation.