Next generation filesystems
Filesystems and indeed the media that they live on have come a long way. Cassettes, cartridges and punchcards didn’t even have a filesystem to speak of: data was just read as a single stream. From there we moved on to directory-less layouts, through to 8.3 filenames, then on to more recognisable features such as permissions and, latterly, journalling. Traditionally, storage has been stratified such that the RAID layer – be it software RAID via MD in the Linux kernel, hardware RAID via a dedicated controller, or so called ‘fake raid’ through drive is broken. There’s no easy way around this that doesn’t involve either adding a new checksumming layer, or combining filesystem, redundancy and checksumming into one. Unsurprisingly, it’s this latter approach that has gained popularity, and it’s one of the defining features of Sun’s ZFS , Microsoft’s barely seen ReFS, and Linux’s Btrfs. Apple’s new APFS filesystem includes checksums for metadata, but not user data.
You’re reading a preview, subscribe to read more.
Start your free 30 days