Layer 3 · Go deeper · 8 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.
Could one 192-vCPU box serve the same 60,000 a second?
Which write loads fit one database primary, and which must shard?
Build-up · 4 steps
Climb the scaling ladder
A stateless web tier scaled out in front of a scaled-up database, the hybrid Stack Overflow and Let’s Encrypt both run. Then the next rungs.
Step 1 of 4
Scale the app tier out
App servers scale out from day one. Keep sessions and uploads off the instances, so any instance can serve any request behind the balancer.
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
“A sale will push your stateless catalogue API to 60,000 requests a second for a few hours. How do you get the capacity?”
Your move. Pick an approach:
Deep dive 2 of 4
Interviewer
“Your startup’s orders database hits 70% CPU at peak, and its instance family has bigger sizes. What do you do?”
Your move. Pick an approach:
Deep dive 3 of 4
Interviewer
“Your commercial database is licensed per core. Do you run many small nodes or a few big ones?”
Your move. Pick an approach:
Deep dive 4 of 4
Interviewer
“p99 is climbing, and the team wants to double the web servers. The primary database sits at 95% CPU. What do you say?”
Your move. Pick an approach: