Practical Elasticsearch — how the inverted index actually finds things, how to size shards before they sink you, how to keep mapping explosions from eating your heap, and how scoring + aggregations behave under real query load.
6 articles · updated regularly
A vendor pushed 47 million docs of untyped JSON into our index. Field count exploded past 12,000. Here's how we crawled out, day by day.
Aggregations are the analytics engine behind Kibana. Know how terms, date_histogram, and cardinality actually execute.
Default BM25 is a strong baseline. Here's how to tune it, combine it with business signals, and not destroy recall.
Dynamic mapping is convenient — and the fastest way to fill your heap with fields you will never query.
Too few shards and you can't scale out; too many and the cluster state eats your heap. The middle is narrower than you think.
From document to token to posting list — the data structure behind sub-millisecond full-text search.