Until recently, the site-search on GitHub Docs was an in-memory solution. While it was a great starting point, we ultimately needed a solution that would scale with our growing needs, so we rewrote it in Elasticsearch. In this blog post, we share how the implementation works and how you can impress users with your site-search by doing the same. How it started Our previous solution couldn’t scale because it required loading all of the records into memory (the Node.js code that Express.js runs).
