InnoDB assumes it's using conventional disks, where random I/O is much more expensive than sequential I/O because of
the time it take to seek to the correct location on disk and wait for the desired part of the disk to rotate under the head.
InnoDB uses its log to convert the random disk I/O to sequential I/O.
Of course, InnoDB does ultmately have to write the changes into data files, because the log file size is fixed.
InnoDB uses a background thread to flush the changes to the data files intelligiently.
The overall log file size is controlled by the innodb_log_file_size and innodb_log_files_in_group , and it's very important for
write performance.
InnoDB uses multiple files as a single circular log.