NTFS
  Treats data like a relational database. 
    Each file is a record.

    Each record contains attributes.
      Such as file's name, size, permissions, ownership.
    
    At least part of each record is stored in the Master file table.
      For small files (< 900 bytes),
        all of file's data may be one of the attributes.

      For large files, 
        attributes hold pointers to clusters holding data.

    Additional attributes including user defined  can be added if "nature"
      of a file changes.

  Almost all structures in NTFS considered files.
    Including Meta-files - files about files - used to record info about files. 
      Lists of files on partition, cluster allocation, free space list, etc.

  Stores data as a b+ tree hiearchical structure.
    Complex to implement but fast access.

    Also minimizes fragmentation to some extent.