Back to roadmap
Module 7 · NoSQL, Search, Graph, ObjectDay 06325 min

Search Engines

Inverted indexes turn 'find me X' into milliseconds.

← PreviousNext →
Day 063

Search Engines

DB
datastore
CDC
service
ES
datastore
Search UI
client
Signal path
Source of truth → search index
DB
datastore
flow
CDC
service
CDC
service
flow
ES
datastore
ES
datastore
flow
Search UI
client
Memory hook

Search Engines: inverted indexes turn 'find me x' into milliseconds

Mental model

match the datastore to the access pattern

Design lens

ES is a derived index — easy to rebuild.

Recall anchors
IndexOps

Why it matters

Search engines invert documents into a posting list per term, allowing fast lookup of all documents containing a word. They handle ranking, faceting, fuzzy matching, and analytics over text.

Deep dive

Sharded across many nodes; querying fan-outs and merges.

Schema mapping decides analyzers and field types.

Source of truth should usually be your DB; ES is a derived index.

Demo / scenario

Add product search to a catalog.

  1. Stream changes from DB → ES via CDC.
  2. Index name, description, attributes with analyzers.
  3. Query: bool filter + match + facets.
  4. Re-index on schema change.

Tradeoffs

  • ES is a derived index — easy to rebuild.
  • Mapping changes can require reindex.
  • Ranking tuning is its own discipline.

Diagram

DB
CDC
ES
Search UI
Source of truth → search index.

Mind map

Check yourself

Loading quiz…

Sources & further reading

← PreviousNext →