HPFS is a "lazy" write system that caches writes to be executed at a later time. Like Unix, it requires proper system shutdown and recovery after a crash may be time consuming. Because of the lazy write characteristic, operations like UNDELETE may not work even in situations in which they would have worked with FAT. Because of the many optimizations, HPFS is reported to be substantially faster than FAT.
HPFS supports File Names up to 254 characters. Case is retained but HPFS is not case sensitive. Code page and foreign language support is built in. HPFS also supports up to 64K of Extended Attributes (EAs) for each file. EAs seem primarily to be used to identify file types and to store file type specific information. Unlike NTFS the long file name is used instead of 8.3 not in addition to, however there is provision for recovering the LFN if the file is set up for 8.3 access.
HPFS allocates disks in units of 1 sector (512 bytes). Fragmentation is purported to be less of a problem than with FAT. HPFS supports disks up to 64Gb although CHKDSK and other important utilities for some versions of OS/2 are limited to 16Gb and some BIOSes will not support HPFS partitions larger than 2.1Gb. Maximum file size is 2.1Gb.
Defragmentation utilities are available. Unlike the file oriented disk organization used in FAT and NTFS, HPFS breaks the disk into 16mb blocks with a directory in the middle of the block. Directories are organized as B-trees with the root directory located in a band at the middle of the disk. A file signature technique is used to enhance error detection. Improved file indexing allows rapid backward referencing within a file.
http://web.archive.org/web/20101018004953/http://www.lesbell.com.au/hpfsfaq.html http://www.edm2.com/0410/hpfs.html http://www.edm2.com/0411/hpfs1.html
Return To Index Copyright 1994-2002 by Donald Kenney.