Making getting this thing reliable is going to maybe a bit harder. Monitoring systems have this sort of "meta" issue of needing to be more reliable than the things they monitor. I'm building and running one as we speak and I had some interesting cases in that regard.
If you can us a SaaS tool, then here is my obligatory plug of my hosted / SaaS https://checklyhq.com.
Only slightly; they just need to have an extremely low likelihood of going down at the same time.
It’s okay if your monitoring is down sporadically, just as long as it doesn’t overlap with the device under test also going down unexpectedly.
A monitoring system that is unreliable risks losing trust. Monitoring systems that aren't trusted tend to be ripped out or ignored.
If it is down too often, people will constantly question whether its silence is an indicator of health.
If it triggers too many false alarms, people will constantly question whether an alarm indicates a real problem. If it gets bad enough, they might start ignoring them.