Design an ad click aggregation service
Company: Meta
Role: Software Engineer
Category: System Design
Difficulty: medium
Interview Round: Onsite
Design a backend service that ingests ad impression and click events and provides aggregated metrics.
Requirements:
- Ingest a high-volume stream of events: {ad_id, campaign_id, user_id(optional), event_type(impression|click), timestamp}.
- Provide near-real-time aggregates (e.g., last 1 min / 5 min / 1 hour) and historical daily aggregates.
- Support queries like: clicks, impressions, CTR by ad_id/campaign_id over a time range.
- Handle duplicate events and late/out-of-order events.
- Be horizontally scalable and fault tolerant.
Out of scope (optional): attribution across devices, fraud detection, complex segmentation.
Overview: This question evaluates a candidate's ability to design scalable, fault-tolerant streaming data ingestion and aggregation systems, including handling high-throughput events, deduplication, late and out-of-order arrivals, and providing near-real-time and historical metrics.