I think this depends on your viewpoint. Every OS needs a mechanism by which it accesses files and data, and that access needs a defined storage and catalogue arrangement which we call a file system. Different OSs do this thier own way and so support different file systems, but every OS must have a file system of some sort if it is to save to and retrieve from a storage medium. Having said that, there's nothing to stop someone binning the standard file system and rewriting their own if they have the skills and a good enough reason to want to do so.
I'm intrigued by your question - are you going somewhere interesting with this?
Re: General File System Question
12 years 1 month ago #14885
Microsoft's new file system WinFS, which was meant to appear in Vista but will now debut in Longhorn, is based on a relational database (actually a derivative of SQL Server). It will be interesting to see how it works.
i found your post quite interesting. affectively you have a number of standard file systems in place like FAT, FAT32, NTFS, and many more. i had at one point of time muddled myself into their origin and their working. in simple terms, if you say FAT16 or FAT32, it only depicts how data is arranged in clusters. if you have more cluster formation capability you have more chance of organizing your data efficiently. lets imagine a specific FAT structure allows you to store 1 MB in a single cluster. if a file is of size 1MB, then it will be stored using just 1 cluster. but say for instance this file is of size 1.5 MB, what happens then? the FAT then splits the file into two halves. the first half has 1MB of the entire file, and the second half has 0.5 MB of the entire file. so you see even the first cluster had been utilized, the second cluster is only half utilized. as a rule you cannot share a cluster between two different files. hence smaller the cluster size you accomodate, lesser file storage efficiency you get out of it. so FAT32 was a better flavour of FAT16 because it allowed bigger files to be stored more efficiently.
but then more you split a file, you encounter problems of fragmentation, lost of file segments, etc etc etc......added to it are the stability issues of all these FAT. NTFS has been recognized as a stable file system.
what i found was that no file system would allow a dynamic cluster size which could increase of decrease according to the file size. this perhaps allow files to be stored in one single stream, and not into clusters. no more issues of fragmentations!!! but then there are so many variables to be take care of while deciding upon a file system
so there you go mate, if you can now design a file system which will allow less fragmentation, provide with more stability, support larger hard drive capacity, allow any size of filenames and be portable across different platforms, say for instance, UNIX, Windows, and Mac for that matter....then i should say you have a vision!!!
so if you are thinking about going that way, let me know. i will be very much interested in it!!!
Re: General File System Question
12 years 1 month ago #14913
There is an ambitious project for a Filesystem, free too, from a charesmatic scientist called Hans Reiser. I'm talking about Reiser4 FS, many of you might be aware of it or it's predecessor -reiserfs-. Though it has no implementation in windows, it introduces some quite revolutionary concepts, so you might be interested to check it out.
A technical tutorial on Reiser4 internals is published in
, less technical intruductions are available in many of Hans Reiser's interviews that you can google for.
Also there's a video of a reiser4 presentation, available