HPFS or High Performance File System, is a file system created specifically for the OS/2operating system to improve upon the limitations of the FAT file system. It was written by Gordon Letwin and others at Microsoft and added to OS/2 version 1.2, at that time still a joint undertaking of Microsoft and IBM.
Among its improvements are
support for mixed case file names, in different code pages;
support for long file names (256 characters as opposed to FAT's 11 characters);
more efficient use of disk space (files are not stored using multiple-sector clusters but on a per-sector basis);
an internal architecture that keeps related items close to each other on the disk volume;
It also can keep 64 KiB of metadata ("Extended attributes") per file.
IBM offers two kind of IFS drivers for this file system: the standard one with a cache limited to 2 MiB, and HPFS386 provided with the server versions of OS/2. HPFS386's cache is limited by the available memory. It's highly tunable by experienced administrators. Thus, HPFS386 is faster, but also IBM has to pay Microsoft for every copy sold.
Because of the Microsoft dependence and the longer disk check times after a crash, IBM ported the journaling file systemJFS to OS/2 as a susbtitute.
There are also third-party drivers for DOS and Linux and official ones for Windows NT.
Design goals and implementation of the new High Performance File System (http://www.2ka.mipt.ru/~alexp/docs/programming/formats/hpfs.pdf) - by Roy Duncan writing for Microsoft Systems Journal (September 1989)
Inside the High Performance File Sytem (http://www.nondot.org/sabre/os/files/FileSystems/HPFS/) - by Dan Bridges writing for Significant Bits magazine (1996)
HPFS not only serves as a way to organize data on random access block storage devices, but is also a software module that translates file-oriented requests from applications programs to device drivers.
The architect of the HPFS started with a clean sheet of paper and designed a file system that can take full advantage of a multitasking environment, and that will be able to cope with any sort of disk device likely to arrive on microcomputers during the next decade.
HPFS uses a new method to represent the location of files that are too large or too fragmented for the Fnode and consist of more than eight runs.