↓ Skip to main content

“The short answer is that we built our own search engine from scratch, in Rust, specifically for the domain of code search.”

Timothy Clemgithub.blog ↗

GitHub spent most of its history failing to make general text search work on code. Elasticsearch took months to index 8 million repositories. Blackbird now covers 45 million repositories, 115 TB of code, and squeezes it into a 25 TB index that re-indexes in about 18 hours. Brute-force grep on the same data would tie up 2,048 cores for 96 seconds per query.