Multilevel inode indexing
= Multilevel inode indexing
An <inode>-based <file system> uses direct pointers for small <files> and successive levels of <indirect block pointers> for larger ones. If an address block contains $q$ pointers, a depth-$k$ region addresses $q^k$ <file data blocks>. With the <inode> resident, an uncached byte access in that region requires $k$ address-block reads and one <file data block> read.