When people first hear the term CDF they intuitively think of data formats in the traditional sense of the word (i.e., messy/convoluted storage of data on disk or tape). CDF is more than just a format. CDF is a ``self-describing'' format for managing data. In addition to the actual data being stored, CDF also stores user-supplied descriptions of the data, known as metadata. This self-describing property allows CDF to be a generic, data-independent format that can store data from a wide variety of disciplines.
In addition to being a self-describing data format, CDF is also a software library. The library routines are callable from C and Fortran and allow the user to randomly access and manage data and metadata without regard to their physical storage. This completely relieves the user of low-level I/O operations allowing more time for data analysis. The actual format used to store the data and metadata is completely transparent to the user.
The term ``CDF'' is also used to refer to the physical files that the CDF library generates. A data set stored using the CDF library is called a ``CDF''.