A web service is running in production. What would you monitor, and why those things?
What they are really testing: Whether you monitor what users feel or what is easy to measure. Leading with CPU and memory is the classic junior tell, because neither tells you whether the service is actually working.
A real interview question
A web service is running in production. What would you monitor, and why those things?
What most people say
drag me
“CPU, memory, and disk usage on the servers.”
These are the easiest things to graph and the least connected to whether users are being served. A service can sit at 30% CPU while returning errors to every request, and this monitoring would show all green.
The follow-ups they ask next
Why p99 rather than average latency?
Averages hide the tail. p99 is the experience of your heaviest and often most valuable users, and tail latency compounds badly when services call each other in a chain.
What makes an alert worth waking someone for?
It is urgent, actionable, and user-visible. If the responder cannot do anything at 3am, it should be a ticket. Every non-actionable page erodes trust in every future page.
What the interviewer is listening for
- Leads with user-facing signals
- Uses percentiles rather than averages
- Distinguishes paging metrics from diagnostic metrics
What sinks the answer
- Starts and stops at CPU and memory
- Alerts on averages
- Cannot say what would page a human
If you genuinely do not know
Say this instead of freezing. Reasoning out loud from what you do know beats silence every single time, and a good interviewer is listening for exactly that.
“First [what users feel: traffic, errors, latency at p95 and p99], then [saturation, because it predicts the next failure]. I alert on [user-visible symptoms tied to an SLO] and keep [CPU and memory for diagnosis, not paging].”
Keep going with observability
Mid
Your team gets 40 alerts a night and mostly ignores them. How do you fix that?
Senior
Define an SLO for a checkout service. What would you pick and why?
Foundation
What is the difference between continuous integration, continuous delivery, and continuous deployment?
Foundation
What is a container, and how is it different from a virtual machine?
Foundation
What does DevOps actually mean, and what problem was it invented to solve?
Foundation
What is infrastructure as code, and why is it better than clicking in a console?
Knowing the answer is not the same as recalling it under pressure
Sign in to send the questions you fumble to spaced recall, so they come back right before you would forget them, and learn the concepts behind them with hands-on labs.
Start free