Skip to content
MTTL, Mean Time To Learn

Beginner3 min read

Once Upon a Time, OpenTelemetry…

OpenTelemetry told as a story: a restaurant too big to see everything at once, a logbook, counters on the wall, a travelling ticket and a shared way of writing things down.

For: engineers and SREs · architects · team managers · business and product · finance, risk and compliance · executives and CIOsPrerequisites: None.

Reading mode

Once upon a time there was a restaurant so big that nobody could see everything at once.

When a guest ordered a beef stew, the order went through a whole chain of stations: order taking, cold prep, the hot line, the sauce station, plating and finally service. Whenever a guest complained about waiting forty-five minutes, the owner asked the same question: where did it get stuck? Nobody could really answer. Everyone swore that things had gone fine at their own station.

So the owner decided to get some clarity and asked the kitchen crew for three things.

The logbook. Each cook writes down what happens at their station: “8:12 pm, out of cream”, “8:15 pm, oven down”. These are logs, the record of what happened.

The counters. Boards on the wall show, in real time, how many dishes go out per hour, the oven temperatures and how many orders are waiting. These are metrics. They don’t tell a story, but they show right away when something starts to drift.

The travelling ticket. This is the cleverest idea. Each order gets a ticket with a unique number. Every station writes down when the order arrived and when it left. At the end, you can read the stew’s whole journey and see, for example, that it sat for twenty-five minutes at the sauce station. That is a trace.

One problem remained. The saucier wrote in Italian, the pastry chef used his own shorthand and the junior cook guessed at the time. Worse, the owner hired outside inspectors to analyse all this and each one demanded their own forms. Switching inspectors meant the whole crew had to relearn how to fill in their notebooks. Everyone grumbled and the restaurant stayed stuck with the same inspector because nobody had the energy to start over.

One day, several big rival restaurants, along with the inspectors themselves, sat down at the same table and said: what if we all wrote things down the same way? That conversation gave birth to OpenTelemetry: a shared language and shared forms for logbooks, counters and tickets, free and open to everyone.

They added one key character: the Collector. It gathers the logbooks, counters and tickets from the whole kitchen, sorts them, removes anything that shouldn’t leave the building (a guest’s credit card number, for instance), then sends everything to the inspector of your choice, or to two inspectors at once if you want to compare them.

The moral of the story

Section titled “The moral of the story”

OpenTelemetry is not a tool that watches your kitchen. It is the shared way of describing what happens inside it. The cooks learn to write things down once and the owner stays free to change inspectors without starting over.

In real life, the restaurant is an application. The stations are the different software services a request passes through and the guest waiting for the stew is you, staring at a page that won’t load.

Going further

Section titled “Going further”