>lots of resources-tracking knobs thanks to cgroups I'm already tracking these at the orchestration level. I don't need my init system to do it. >support for

>lots of resources-tracking knobs thanks to cgroups I'm already tracking these at the orchestration level. I don't need my init system to do it. >support for read-only root filesystem Nice, I guess, but not sure why it's a particular win for cloud. My stuff is already stateless by design. I don't care if anything gets written to the root volume or not, since it will go away when the stateless vm goes away anyway >many batteries included (no need to ship ntp on VMs thanks timesyncd, journald can replace syslogd for basic usage, networkd tries to do the right thing without any manual setup when run in a VM, etc. Well, I could strip out NTP thanks to the systemd SNTP client, but I could also just use NTP and not use the systemd SNTP. The ephemeral nature of cloud makes journald absolutely useless, unless you just don't care about logs. (Why don't you care about logs?) Restarting services, containers, and networkd are nice, but I think the other stuff is probably reaching in a lot of ways.