Design and scale a Yelp-like platform
Overview
This question evaluates system design and engineering skills including data modeling, indexing, geospatial and full-text search, caching, replication and sharding for a high-scale reviews platform, together with key generation, storage layout, redirects, rate limiting and analytics for a URL-shortening service, probing the ability to balance scalability, consistency, availability, and operational complexity. It falls under the System Design / Distributed Systems / Databases domain and is commonly asked to assess architectural reasoning and trade-off analysis, testing both conceptual high-level design and practical implementation considerations such as performance, fault tolerance, and operational constraints.
System Design: Yelp-like Local-Business Reviews Platform + URL Shortener
You are designing two systems:
- A Yelp-like local-business reviews platform that supports millions of users, businesses with geocoordinates, categories, reviews, ratings, photos, and check-ins.
- A high-scale URL-shortening service.
Part A — Yelp-like Reviews Platform
Define and discuss:
-
Core data model
- Entities: users, businesses (with geocoordinates), categories, reviews, ratings, photos, check-ins.
- Indexes for common queries.
-
Scaling reads
- Replication, caches, CDN, search indexes.
-
Scaling writes
- Sharding, queues, consistency (read-your-writes), backfills/migrations.
-
Search and geospatial queries
- Full-text, category filters, distance/radius queries, sorting.
-
Cache invalidation and performance
- What to cache, when/how to invalidate, pagination, hot keys, tail latency.
Part B — URL-Shortening Service
Outline a design covering:
- Key/ID generation and collision handling (sequential vs random, length analysis).
- Storage layout and data model.
- Redirects, HTTP status codes, and CDN usage.
- Rate limiting and abuse prevention.
- Analytics and aggregation.
- TTL/expiration, deletion, and alias reuse policy.
- Custom aliases and namespace management.
- Strategies to scale high read/write traffic (multi-region, cache, sharding).
Solution
Login required