Explain like I'm five · SRE

What is SRE?

Site Reliability Engineering, told as a story about the people who keep an amusement park open every single day.

watching even at nightthis park never closes — not one day a year

This park never closes — not one day a year.Guests come day and night. The control room never goes dark.

brokenthe line keeps growingno ride is perfectthat one's fine though?the goal isn't never breaking — it's breaking without guests noticing

But every ride breaks, eventually.The goal isn't never breaking. It's breaking in a way guests never notice.

SRE = the people who keep trouble invisible to guestswatch the gauges, automate what repeats, answer the page, write it down after

SRE is the people who keep trouble invisible to guests.Watch the gauges, automate what repeats, answer the page, write it down after.

Watch the gaugescatch it before it breaks
Automate the repeatsno one can watch it all
Answer the pageeven at night
Write it down afterwithout blame
PARK MAP12 rides runningcontrol room watches allone breaks → mechanic respondsstill stuck → switch to sparespare generatorno one watches it alone — gauges, response, and a spare all work together

No one watches it alone — gauges, response, and a spare all work together.When a ride breaks, a mechanic responds. If that's not enough, the spare generator takes over.

never 100%chasing perfect costs without endOUR PROMISEtarget: 99.9%free to move within itcross it, everyone pausesdecide in advance how much to promiseand automation takes over the repeats — but a person still makes the final call

There's no such thing as perfect uptime. Instead, promise in advance.Decide how much you'll promise, and move freely within it. The final call always belongs to a person.

In one breath

SRE = keeping the park open every day — not by chasing perfection, but by promising an uptime in advance and keeping that promise with gauges, automation, paging, and postmortems.

A discipline Google created to apply software-engineering methods to operations. The core idea isn't 100% uptime — it's meeting a target set by an SLO, automating repetitive work, and learning from every incident.

When grown-ups say it

Site Reliability Engineering
The whole job of keeping the park running. Observability, automation, on-call, incident response, and learning — all of it.
Uptime
How long the park stayed open. 100% doesn't exist in reality — so you set a target instead.
SLO
The promise line we set ourselves. Cross it, and everyone pauses. → the target line we set ourselves
Automation
The machine that runs instead of a person. When the line grows, it calls in more staff on its own. → calling in more staff
On-call
Whoever's holding the pager this week. Answers the page, even at night. → whoever's holding the pager
Incident response
The mechanic rushing over. Stops a small break from becoming a big one.
Postmortem
The record written without blame. So the same thing doesn't happen twice. → the record everyone writes without blame
Golden signals
The four gauges on the control-room wall. Latency, traffic, errors, saturation. → the four gauges on the control-room wall
Next story
Today's Ride Log →
Want another term as a picture book? Request →