Database log

redo log

redo log： Redo log , Used to record changes in transaction operations , Ensure transaction persistence .redo log Start recording after the transaction starts （ It's not a business commit It's not recorded until the end of the year , Because there may be many changes in the whole transaction , If in commit Only when you write redo log Words , It may take a long time . It's business begin It was recorded after the beginning , With the modification of data during transaction execution , Write in real time redo log）, Whether the transaction is committed or not is recorded , When an exception occurs （ Such as power failure during data persistence ）,InnoDB Will use redo log Back to the time before the power down , Ensure data integrity .

innodb_log_buffer_size The default is 16M, Namely redo log Size of buffer , It starts with the transaction , Start writing redolog, If the transaction is big , In order to avoid spending too much disk during transaction execution IO, You can set a larger redo log cache , Save disk IO. There is a time to refresh the disk . When the time comes, it costs disk IO, If buffer The larger , Will reach the refresh time more slowly , More efficient .

InnoDB Modify operation data , Not directly modify the data on the disk , Actually, it's just a modification Buffer Pool Data in .InnoDB Always put it first Buffer Pool Data changes in are recorded to redo log in , Used for data recovery after crash . Priority record redo log, Then find a time to slowly Buffer Pool The dirty data in is flushed to disk .

innodb_log_group_home_dir Two files in the specified directory ：ib_logfile0,ib_logfile1, This file is called the redo log .
buffer pool Buffer pool ：（B+ Tree index , Adaptive hashing ）
effect ： Speed up reading and speed up writing , Direct manipulation data page, Write redo log Even if the modification is completed , There are special threads to do buffer pool Medium dirty page Write to disk .

undo log Rollback log

undo log： Rollback log , Save a version of the data before the transaction , Used for rollback operation during transaction execution , At the same time, it also realizes multi version concurrency control （MVCC） The key technology of read down operation .

undolog The main function of rollback log ：

• 1. Rollback when a transaction has an error rollback, Rollback log （ When the data is updated , Save the data before the change in the rollback log , The purpose is to roll back the transaction in case of error or when we roll back manually , The initial data can be found in the rollback log ）
• 2. Provides MVCC Non locked reading （ Read the snapshot ）