Design video platform and catalog system
Overview
This System Design question evaluates proficiency in large-scale architecture, API and storage design, data modeling, scalability and consistency trade-offs, ingestion and indexing pipelines, CDN and caching strategies, and operational concerns such as reliability, observability, security, and multi-region deployments across video-sharing and marketplace catalog domains. It is commonly asked in technical interviews to test architectural judgment and trade-off reasoning, emphasizing high-level practical application and operational planning rather than low-level implementation details.
System Design: Two Independent Systems
You are asked to design two large-scale systems. For each system, provide:
- Scope and key requirements
- External APIs (read/write) and internal service interfaces
- Storage choices and data modeling
- Scalability strategy and high-level architecture
- Consistency model and data flow
- Reliability, availability, and observability plans
Where relevant, estimate traffic/storage, and discuss cost and security trade-offs.
1) Video Sharing Platform (YouTube-like)
Requirements:
- Video upload, validation, and asynchronous transcoding into multiple bitrates and codecs
- Storage tiering (hot/warm/cold), CDN delivery with adaptive streaming (HLS/DASH)
- Metadata and search, user actions (watch, like, comment)
- Recommendations interface (feature and event APIs to interact with a separate recs system)
- Estimate traffic/storage; design ingestion pipeline, thumbnail generation, indexing
- Caching (metadata, watch pages), rate limiting and abuse controls
- Multi-region availability, encryption/DRM, and cost trade-offs
Deliver:
- Architecture, APIs, queues, storage choices
- Consistency choices for user actions and metadata vs search
- Scalability strategy for upload, transcode, playback, and indexing
- Reliability and observability plan
2) Marketplace Catalog System
Requirements:
- Large catalog with categories, attributes, products, and product variants (SKUs)
- Item ingestion (batch/stream), versioning, deduplication, and moderation workflows
- Services for search/browse indexing with localization and incremental updates
- Read/write APIs, caching, and consistency model between the source of truth (SQL/NoSQL) and the search index
- Address schema evolution, multi-tenant isolation, and backfill/reprocessing strategies
Deliver:
- Domain model and storage choices
- Ingestion and moderation flows
- Indexing pipeline and cache strategy
- Consistency, availability, and observability plans
Solution
Login required