The alerting problem
By Vimal Bhaya, Founder and Lead Systems Architect · Last reviewed 3 September 2026
Freezer alarms fail for reasons that have nothing to do with the sensor. They watch the air instead of the food, they cannot tell a defrost from a breakdown, and they often go to one phone that nobody is holding.
This is a five-part tutorial on how to fix that. Each piece is a full article, with the readings from three published failures. Read them in order, or jump to the one that matches the problem you have today.
The 5 parts, in order
By the end you will know how to tell a harmless spike from a real failure, why a probe in the air lies to you, who should get woken, and what an alert has to say at three in the morning.
-
1. Why temperature alarms cry wolf
Every alarm threshold an operator might pick, replayed against 5,179 real readings. There is no setting that gives you both a quiet week and useful warning. This is the problem the rest of the tutorial solves.
-
2. Threshold vs rate-of-change alerting
The steepest hour in our whole record was a freezer working perfectly. The real emergency crept up at less than a degree an hour. Why the shape of the line tells you which one you have.
-
3. Glycol buffers and thermobuffers
Your groceries survive the drive home because food changes temperature slowly. Air does not. What a sealed bottle of antifreeze around the probe actually fixes, and when it is the wrong choice.
-
4. Alerting more than one person
A 3am alert to one sleeping phone is the same as no alert. How escalation from email to SMS to a ringing phone works, and why it is set per freezer rather than once for the whole site.
-
5. What a good alert looks like at 3am
Six things an alert has to say before it is worth waking somebody. Built from a real Sunday morning alert that bought two working days of runway over a holiday weekend.
How to use this
Read them in order if you are starting from scratch. That is the usual path. If you already know which piece you need, jump to it from the list above.
Every figure in the series traces to a published case-study export. The three failures behind it are the Philadelphia freezer, the dual-system walk-in, and the Mississauga reefer.