What a Filesystem Is
conceptEveryone has saved a file. It feels like the file goes "onto the disk"—a thing, placed in a place. But a disk—a hard drive, an SSD, a USB stick— offers nothing so friendly. It stores only blocks: millions of small, numbered, same-sized slots that each hold some bytes. A disk has no idea what a file is. It can do exactly two things: read slot number so-and-so, and write slot number so-and-so.
A filesystem is the bookkeeping software that turns those anonymous slots into the world you actually use: files with names and sizes, folders inside folders, a record of who is allowed to read what.
So the correction worth carrying from here on: your files are not really "on the disk." The disk holds only blocks; the files live in the ledger. Lose the ledger and every block is still there—and means nothing.
You have been using filesystems for years without naming them:
- "Formatting" a drive means writing a fresh, empty ledger onto it. Nothing about the hardware changes—the drive just gets a new, blank book of records.
- Every device picked a ledger design: USB sticks usually carry FAT32 or exFAT, Windows uses NTFS, Macs use APFS, Linux most often uses ext4. When a Mac can't write to a Windows drive, that isn't a hardware problem—the two systems keep their records differently, and one side can't read the other's book.
- "The format erased my files" means: new ledger, old entries gone. The blocks underneath often still hold the old bytes for a while—which is exactly what data-recovery tools go digging for.
The words that get tangled together, untangled:
- The disk is hardware. It holds blocks. That is all it does.
- A partition is a fenced-off region of a disk—a way to split one physical disk into sections, each of which gets its own filesystem.
- The filesystem is the ledger written inside a partition (or a whole disk): the software layer that invents files.
- The operating system is the reader and writer of that ledger—it contains the code that understands each format.
Stacked in order: disk → partition → filesystem → files. Each layer adds meaning the layer below doesn't have. Almost everything in storage that seems confusing becomes clear the moment you ask "which layer is this about?"—and ZFS, where this site is headed, is interesting precisely because it redraws these layer boundaries.
Every filesystem ever built is a set of answers to the same four problems:
- Allocation — which blocks are free, and which belong to which file?
- Naming — how does
/photos/cat.jpgbecome a list of block numbers? - Consistency — what happens if the power dies while the ledger is half-updated? A ledger that lies is worse than no ledger.
- Speed — disks are slow and memory is fast; what gets cached, and when is it safe to say "done"?
Different filesystems are different trade-offs among the four. And the history of filesystems—the next node—is mostly the history of problem 3: each generation discovered a new way the ledger could end up wrong, and built a new defense. The most recent generation, the one ZFS opened, added a fifth problem the first four decades quietly ignored: what if the disk itself lies?