LEDs also cause interference with WiFi and other radio signals. Found out after putting a LED bulb in my smart garage door opener.. which immediately went offline.
Worked on a team at BofA, our application would handle 800 million events per day. The logic we had for retry and failure was solid. We also had redundancy across multiple DCs. I think we processed like 99.9999999% of all events successfully. (Basically all of them, last year we lost about 2,000 events total) I didn’t find it very stressful at all. We build in JMX Utica for our production support teams be able to handle practically anything they would need to.