Layer 3 · Go deeper · 6 min

Go deeper

Size it, build it, then make the calls an interviewer will push you on.

Build-up · 5 steps

A redelivery after a crash

One message, two workers, an SQS visibility timeout of 30 seconds. Time runs left to right; drawn to show order, roughly to scale.

Step 1 of 5

The producer sends m1 at 0 s. The queue stores it durably before saying ok.

4 interview moments

The interviewer pushes on the hard parts. Pick an approach and see how it holds up.

Deep dive 1 of 4
Interviewer
“Checkout calls the invoice service and waits for it. On sale days, slow invoicing times out checkout. What do you change?”
Your move. Pick an approach:
Deep dive 2 of 4
Interviewer
“Report jobs take 2 to 8 minutes, and the SQS visibility timeout is still the default 30 seconds. What do you set?”
Your move. Pick an approach:
Deep dive 3 of 4
Interviewer
“OTP SMS messages are arriving late, and the worker fleet isn’t scaling up. Which signal should drive scaling and alerts?”
Your move. Pick an approach:
Deep dive 4 of 4
Interviewer
“A one-hour outage queued 4.32 million messages. The fleet has 560 messages a second of spare capacity. What do you do?”
Your move. Pick an approach:

Build-up · 5 steps

A worker fleet processes at most 1,200 messages a second. Normal traffic is 400; a flash sale sends 2,000 a second from minute 5 to minute 15.

Step 1 of 5

Arrivals at 400 a second sit well under the fleet’s 1,200, so the queue stays empty.