It’s useful from the standpoint if someone calls and says they were down 3 times in the past two weeks, we can look and confirm whether our equipment was down or it was something else. We can also see changes to response times, thus potential load, isolated bottleneck or interference issues, see persistent or inconsistent packet loss, etc, on every device in our network, allowing me to trace problems that occurred up to 1 month previous through the network if necessary. It’s a tool used in conjunction with the other graphs, but it’s my first stop when there’s an outage reported and in the first 3 stops when people complain about speed issues.
I have something for troubleshooting known problematic connections, but I can’t point that at every device we monitor. I do like your example above, if I can run a cronjob targeting problem connections and get librenms to display the data, that will be much better than what I’m using.