Pasted 1 month, 1 week ago — Expires in 325 days
I was listening to BSD Now and you talk about ZFS, and have a question for you. I have a 1 T SSD with 200 G data on it, and it’s 1/5 full. Does this make sense to you? Hold on. Why NOT store 5 copies of everything? Why is having a drive with only 1 copy of files that is mostly empty such a ‘sacred cow’? I was thinking that it makes sense to FILL the drive with duplicates at the operating system level. So, a drive might start at 1/10 full, with 10 copies of all files, then drop to 5 copies of everything, to 3 copied of everything, to 2 copies, then to 1 copy. What about fragmentation? Well, thought about that too - you could either over allocate (and assume that future copies of files would be larger than the existing copy), or you could look at file sizes as a means of picking what should be over written. For instance, I could have 4 copies of a movie at 4 gig each. You could allocate 4.2 gig per copy. And when I need to upload a VERY high resolution copy of the same movie, you could replace them with 2 copies at 4.2x2 each. OR, if I had the same 4 copies of a 4 gig movie, and uploaded a new movie, I could replace 2 of those with the new movie. You could - maybe- hide these from the OS, so that if you got crypto ware, they would remain intact. Or, simply allow reversion back to previous copy. Anyway, so why is keeping empty space so important? Seems like you don’t need to.