With the "sleep 60" command, the script pings at 63 seconds, then 65 seconds later, then 62 seconds later, and so on. After 24 hours, the whole script has drifted forward by 2 or 3 minutes.

For example, if I have to run a command on 500 machines it would be nice to be able to run more than one at a time instead of running them in order as they would reduce the runtime by quite a bit.