Linux

But you are doing the work the computer should do by scripting your own startup process. Also, it will process your rc.local sequentially whereas systemd does things in parallel. If you have 5 different custom services that need network, your approach would have them started one after another. Systemd would wait for network access and then start them all in parallel. If one of those hangs, the others will still start in a few seconds (unless they depend on the hanging service) and the boot process will still continue.

Also, what about if some service fails? systemd can restart them automatically, you have commands to see at a glance whether your desired services are all running (i.e. the system is in your desired state), it manages the log outputs for each service, etc. etc. .... it's a huge comfort win and once you've written a few units, you won't have to look everything up all the time.

[Unit]
Description=My service
After=network-online.target

[Service]
ExecStart=/usr/local/bin/myservice -d

[Install]
WantedBy=multi-user.target

Put this in /etc/systemd/system/myservice.service, run systemctl daemon-reload followed by systemctl enable myservice and Bob's your mother's brother. Optionally, start it directly using systemctl start myservice. (On most systems, service myservice start will work, too.) It doesn't get any easier than that.

And, if you start to automate your system's configuration(s) using e.g. Ansible, it's far easier to just place a few files in the filesystem and run a few commands than to modify the rc.local in an automated fashion without breaking something.

While I don't really like the one-tool-for-everything approach with systemd and its various additional features (timedated, resolvd, etc.), I do like the main feature.