sam_14332 wrote:Designing a good filesystem is hard, take my word for it.
Absolutely NOT, implementing a filesystem is something you could do for breakfast.
I've popped up your thesis - it goes rambling on a full 7 pages on your filesystem design, and even in the small bit I've read I've got some doubts about wether things will actually work. As far as I can see you did consider a few things, but haven't at all stresstested it on performance. As a master student, I would IMHO expect that you at least give some sort of proof that your FS works that well as you just kindof claimed it does.
Besides, that your lunch came for free isn't really cause that someone else's lunch came for free. I think that a rather large number of people here would not be able to fully understand what's in there.
And given that the review discussions on SFS (which is designed to be the simplest FS) have been far longer and thorough than those 7 pages, I expect that you could spend a thesis on its own really discussing a full-blown FS.
Hence I tend to agree with the (popular) opinion that designing a good FS is really an expert's job. Implementing a complex filesystems is of course a lesser burden, but still not insignificant.