Layer 3 · Go deeper · 7 min
Go deeper
Size it, build it, then make the calls an interviewer will push you on.
Back of the envelope
Size it up
Tap through each step of the arithmetic.
When should a 99.9% checkout SLO page someone?
Errors hit 6% for 20 minutes. When does the page fire and clear?
Build-up · 6 steps
Follow one slow checkout from page to log
Checkout’s SLO: 99.9% of requests within 1.5 s, and its p99 is normally about 1 s. Then it doubles. One hop per step.
Step 1 of 6
The page fires at 19:57
About 4% of checkouts exceed 1.5 s: a burn rate of 4% ÷ 0.1% = 40. The 1-hour ratio needs ≈ 22 minutes to cross 1.44%, so trouble began around 19:35.
4 interview moments
You’re the candidate
The interviewer pushes on the hard parts. Pick an approach and see how it holds up.
Deep dive 1 of 4
Interviewer
“Your fleet dashboard shows checkout’s p99 as the average of each pod’s p99. How should it compute the fleet p99?”
Your move. Pick an approach:
Deep dive 2 of 4
Interviewer
“One customer says checkout is slow. A teammate adds user_id as a label on the latency histogram to debug it. Approve?”
The histogram has 14 series per label set: 20 routes × 5 status classes × 50 pods gives 70,000 series today.
Your move. Pick an approach:
Deep dive 3 of 4
Interviewer
“Checkout serves 5,000 requests a second with 25 spans each. Which traces do you keep?”
Your move. Pick an approach:
Deep dive 4 of 4
Interviewer
“Checkout has a 99.9% SLO over 30 days. What should page the on-call engineer?”
Your move. Pick an approach: