...
As described in the previous chapter, when the redo log is recorded during the transaction processing, the checkpoint step for the persistence of the memory DB, and the disk I/O for transaction processing of the disk DB are configured on the same disk, a disk bottleneck occurs. It is recommended to configure the redo log and each area as a physically separated disk in order to minimize the performance downgrade.
Disk I/O
As shown in the figure above, if the redo log record, checkpoint, and disk I/O by the buffer manager are concentrated on one disk, the disk bottleneck cannot be avoided. Therefore, it is recommended for the user to refer to this document for efficient disk configuration.
...