Track 3 · Booking and Search
Search From the Ground Up
Why LIKE '%word%' dies at scale, what an inverted index really is, how text becomes tokens, how shards answer one query together, and a lab you can run on your laptop. Written around Flipkart, Zomato, IRCTC and Swiggy style search boxes.
Guide24 min read
On this page
- The night the dish search melted
- The simple version
- Analysis: how text becomes tokens
- Documents, mappings and two kinds of field
- Why this result is ranked first
- Inside one shard: segments, refresh and the translog
- Shards, replicas and one query on many machines
- Where Elasticsearch fits, and where it does not
- Lab: run it on your laptop
- The war room: the update storm
- When to use what
- What a strong interview answer sounds like