The Problem
A news aggregator is a feed delivery system. The product is about collecting article metadata from many publishers and serving it quickly by region.
Requirements
- Aggregate articles from many publishers
- Paginate the feed
- Redirect users to the publisher site on click
Core Entities
- Article
- Publisher
- User region
Only metadata and redirect URLs belong in the system. The full article stays on the publisher’s site.
Ingestion and Read Paths
The architecture has two flows:
- Ingestion: poll publisher feeds, extract metadata, and store article records
- Read: serve the feed by region with a sorted query or cache hit
Pagination
Cursor pagination is better than offset pagination because it avoids duplicates and skipped rows when new articles arrive between requests.
Caching and Regional Delivery
Regional Redis caches make the feed fast enough for large traffic spikes. A short TTL is fine because slight staleness is acceptable for news.
Final Shape
The clean interview answer is: poll publishers, store article metadata, serve a regional feed with cursor pagination, and use cache layers to keep latency low.