Ten Times the Traffic
You have six weeks to get a platform ready for a national TV moment projected at ten times its peak traffic. One production environment, a database everyone fears, no recent load test - and leadership asking one question: will it hold.
the situation
You have just been brought in to own platform readiness at a consumer company that signed a marquee partnership: a national TV moment goes live in six weeks, and marketing projects ten times normal peak traffic across a 48-hour window. The platform has never seen more than twice its usual load. There is one production environment, a relational database that everyone is quietly afraid of, and the last load test anyone can point to is years old. The engineers who built the system are helpful but stretched, and nobody can tell you where it actually breaks. Leadership is not asking for architecture diagrams. They are asking one question, and they want an honest answer: will it hold.
what this interview covers
- Finding the true bottleneck
- Load testing against reality
- The incident, live
- Degradation and capacity design
- Cost, the runbook, and what remains
How the interview works
One scenario, 45 minutes, hard stop. The interviewer works through the situation with you - design, incident, trade-offs - and will not confirm or correct anything until the debrief. There are more questions in the arc than most people reach; how far you get matters less than how well you reason. Saying you do not know and moving on beats bluffing. When time runs out, you are graded on everything you covered.
about this interview
About
A consumer company has just bought its way onto national TV, and the traffic projection is ten times anything the platform has ever survived. You have six weeks, one production environment, a relational database the whole team tiptoes around, and a leadership team that wants one honest answer: will it hold. This interview runs the whole arc - modeling the spike, load testing against reality, a mid-preparation incident of your own making, the degradation and capacity design, the money conversation, the day-of war room, and what you leave behind when the lights go off.
The interviewer plays a senior engineer running a real screen. They will push on your reasoning and ask why, and that pressure is the format, not a verdict. What they will not do is confirm, correct, or hint at how you are doing while the clock runs. Any stack earns credit here - the questions are about capacity, failure, and judgment, not about naming a particular vendor's products. If you do not know something, say so and keep moving; bluffing reads worse than honesty.
You get 45 minutes and the stop is hard: when time runs out, the interview ends wherever you are, exactly like a real screen. After that comes the debrief real interviews never give you - what each answer showed, what a strong answer looks like, and an overall hire signal based on everything you covered.