Datasets stored on the databases can be accessed in raw mode, i.e by writing and reading information to and from byte streams. This is often more efficient and meets programming requirements better than accessing data with the DB manager. A prerequisite for using raw mode is some knowledge about UNIX file input/output.