Interactive essays on the machinery of software. Every figure is something you can drag, tune, and break; the articles work best when you play with all of them.

• The Elasticsearch essays Anatomy of a search

  How a document becomes five data structures, how text is normalized, how the inverted index answers without reading, how queries compose and forgive typos, and how BM25 ranks what comes back.

• The Elasticsearch essays Segments: how writes really work

  Why Elasticsearch never modifies a file: buffers, refreshes, immutable segments, tombstoned deletes, and the constant background merging, as a live simulation you operate.

• The Elasticsearch essays One machine from many

  Distributed search in two phases, shards and replicas, node failure, and the quorum election that prevents split-brain.

• The Elasticsearch essays Numbers at scale

  Doc values, aggregations, and honest approximations: summarizing millions of documents, and counting distinct users in 256 bytes.

• The Elasticsearch essays The metrics engine

  25 bytes per data point down to 3.75, skip-before-decode queries, a rate() that survives counter resets, and an ingest path without ceremony.

• The Elasticsearch essays The life of a log

  From raw line to structured document to cheap old age: ingest pipelines, data streams, and the lifecycle tiers that made keeping years of logs affordable.

• The Elasticsearch essays The vector database

  Meaning as geometry: embeddings, HNSW navigation, filtered vector search, BBQ quantization, and hybrid ranking with reciprocal rank fusion.

• The Elasticsearch essays ES|QL: one language for the whole machine

  From pipes to plan trees to pages of blocks: the parser, the optimizer, the physical planner, and the vectorized compute engine.

• The Elasticsearch essays The machine, whole

  Document store, columnar store, vector store: three specialized engines wearing one coat, and the single query that uses all of them at once.