MongoDB uses a journaling system to ensure data consistency and availability. When data is written to the database, it is first written to a private view in memory, which is not directly connected to the underlying file on disk. The change is then written to the journal file, creating a description of the changes made. The journal appends each new change description, allowing the system to recover from crashes by replaying the changes from the journal back onto the shared view. Finally, the shared view is flushed to disk at regular intervals, ensuring that data consistency is maintained.