Explain like I'm five · SRE

What is Toil?

Toil, told as a story about an operator who copies the exact same report by hand every morning.

TODAY'S REPORTsame as yesterdaycopying it by hand, again40 minutes in...it just keeps piling upevery morning, the operator copies the exact same report by hand

Every morning, the operator copies the exact same report by hand.The same content as yesterday, copied by hand again today.

no time left to look at anything newwhen repetitive manual work piles up, it swallows all the time a person has

When repetitive manual work piles up, it swallows all the time a person has.The more paper piles up, the less time is left for anything new.

do I really have to do this by hand every time?the machine does it insteadI'll go look at something newlet the machine handle what repeats, and spend people's time on what's newkeep asking: do I really have to do this by hand every time?

Let the machine handle what repeats, and spend people's time on what's new.Keep asking — 'do I really have to do this by hand every time?'

Find repetitive, low-value workand whether it can be automated
Measure how much time it takesnumbers, not a guess
Decide whether to automateweighed against the cost
Can't automate? Trim it downeven if not all the way
BEFORE AUTOMATINGtoil time: a lotimprovement time: almost nonetoilAFTER AUTOMATINGtoil time: littleimprovement time: a lotimprovementthe more you automate, the balance tips from toil toward improvement

The more you automate, the balance tips from toil toward improvement.Time is limited — less on one side means more on the other.

this only happens once a year — should I even automate it?WEIGH IT UPtime to build automation: 3 daysdoing it by hand: 10 min a year-> just do it by handnot every repeat is bad — automating something rare can cost more than just doing it

Not every repeat is bad — automating something rare can cost more than just doing it.You have to weigh the cost to build it against the time doing it by hand actually takes.

In one breath

Toil = repetitive, low-value manual work. Leave it unautomated, and it swallows all the time a person has.

A concept defined by Google SRE for operational work that's repetitive, automatable, and has little long-term value. Teams measure the toil ratio to manage it, and decide whether to automate by weighing the cost against the time it would save. The time freed up as toil shrinks often leads to bigger automation, like orchestration.

When grown-ups say it

Toil
Repetitive, low-value manual work. The more you automate, the less of it there is. → the people who keep the park open
Automation vs. manual
The line between what the machine does and what a person does. Repetition goes to the machine; judgment stays with the person.
Measuring toil
Timing how much time a task actually costs. Without measuring it, you can't even tell if it shrank.
Separating it from valuable work
Drawing a line against work that solves something new. Toil solves the exact same problem, over and over.
Deciding whether to invest in automation
Weighing the cost to build against the time it saves. Something rare enough might not be worth it.
Operational burden
The overall weight that builds up from accumulated toil. Pile it up, and there's no time left for anything new.
Orchestration
Once automation grows, coordinating many of them together. The next stage after automation starts cutting toil.
Runbook
The notebook where the manual steps were written down. Automation usually starts by turning this notebook into code.
Next story
The People Who Keep an Amusement Park Open →
Want another term as a picture book? Request →