An honest benchmark would include the non-snap startup times for each of their test platforms. That said, the performance issues aren’t as bad as the fact that

An honest benchmark would include the non-snap startup times for each of their test platforms. That said, the performance issues aren’t as bad as the fact that it’s unusable on machines with NFS home directories mounted with root_squash (as documented [0] by someone at another institution). We’re a University CS dept with a few hundred lab machines and have been using Ubuntu for the past decade or so, but fighting against the increasing snapification every LTS is frustrating. Next LTS we’ll probably just switch to another distro. Most of our users don’t really care which distro they use - they just want Linux with NFS home, a browser, JetBrains, VS Code and a Bash shell with access to our standard compiler and tool collection Our HPC team recently switched from Centos to Alma - though I’m more inclined to stay with a Debian derivative (or more likely Debian itself). [0] https://utcc.utoronto.ca/~cks/space/blog/linux/SnapsVersusNFSHomedirs https://utcc.utoronto.ca/~cks/space/blog/linux/SnapsVersusNF...