Comment Re:Who actually uses ZFS? (Score 1) 279
My needs are much more modest. Mine is a home environment and which includes a backup file server which I migrated from MDRAID to ZFS (RAIDZ2) a couple years ago. I also have a remote server colocated at my son's place where I keep a mirror of my home (backup) server. I used to use `rsync` to mirror data from the local backup set to the remote server. There were two problems with that. First, `rsync` needed to examine both data sets to determine what had changed. Second, `rsync` cannot preserve hard links when copying large data sets. My disk management strategy included deduplicating some files via hard links. ZFS solved both issues. Sending incremental backups reduces network usage to only change data (as opposed to identifying changes first, then sending.) Second, since ZFS send/receive duplicates filesystems instead of files, hard links are preserved when transferring to the destination. In addition to these benefits, I appreciate the extra processing that preserves file integrity in the data sets (and yes, both servers are populated with ECC RAM and are on server motherboards/systems.)
My second use for ZFS was on a then new XPS-13. I could not get Killer WiFi working with Debian Stretch. Debian Buster was still "Testing" and not having ever used Testing, I installed to ZFS on Root so that I c,mirrould roll back should a package upgrade cause problems. It turned out that the only time I used this was when a kernel upgrade required ZFS utilities not yet in the repos. There are other ways to manage this but the easiest for me was to roll back. Buster is now Stable and when I upgraded the SSD on my laptop I reinstalled using ZFS native encryption.
ZFS has been much more than a buzzword for me. I understand Torvald's refusal to include it in the kernel until (and when) the license issues are resolved. I'm a little disappointed in him dissing something that he apparently knows little about.