WAFL - Write Anywhere File Layout

In many respects, WAFL's disk format is similar to that of other UNIX file systems such as the Berkeley Fast File System [MJLF84] and TransArc's Episode [CAK⁺92] file system. For example:

• WAFL is block based, using 4 KB blocks with no fragments.
• WAFL uses inodes to describe its files.
• Directories are specially formatted files.

Like Episode, WAFL uses files to store meta-data. WAFL's two most important meta-data files are the inode file (which contains all inodes) and the free block bitmap file. Keeping meta-data in files allows meta-data blocks to be written anywhere on disk. WAFL has complete flexibility in its write allocation policies because no blocks are permanently assigned to fixed disk locations as they are in the Berkeley Fast File System (FFS). The only exception to the write anywhere policy is that one inode (in WAFL's case the inode describing the inode file) must be written in a fixed location in order to enable the system to find everything else. Naturally, this inode is written redundantly.