Design Rolling-Window Ad Frequency Capping
Company: Netflix
Role: Software Engineer
Category: System Design
Difficulty: medium
Interview Round: Onsite
Design a frequency-capping service for an advertising platform.
The ad platform receives ad requests for users and must decide whether a candidate ad, ad group, or campaign can be shown. Each cap has the form: a user may receive at most `N` impressions for a given ad scope during the last `W` units of time. The window is rolling, not calendar-based.
Your design should cover:
- the read path used when deciding whether to show an ad,
- the write path used after an ad impression is served,
- data models for cap definitions and impression history,
- how to enforce caps correctly under high request volume,
- how to handle distributed-system issues such as race conditions, duplicate requests, failures, and data retention.
Quick Answer: This question evaluates system-design skills for building scalable, low-latency services that enforce time-based rolling-window state, including data modeling for impression history, read/write paths, concurrency control, and fault tolerance; category: System Design.