His existing monitoring setup has snmp traps n rsyslog involved somehow? It didnt detect the outage soon enough for them to prevent the issue from escalating into...whatever it is now...

I mean...end user complaints is the cheapest kind of monitoring solution but maybe its not politically acceptable?

You could totally do redundancy and failovers and stuff. And perhaps automate some of that. And monitoring is there to detect the "down" to trigger some of these?