Which Cypher statement creates a uniqueness *and* an index in one go?

CREATE CONSTRAINT person_email_unique FOR (p:Person) REQUIRE p.email IS UNIQUE

CREATE INDEX person_email FOR (p:Person) ON (p.email)

CREATE CONSTRAINT person_email_unique FOR (p:Person) REQUIRE p.email IS UNIQUE

CREATE UNIQUE (p:Person) ASSERT p.email

ALTER (p:Person) ADD UNIQUE email

Your query is `MATCH (u:User) WHERE u.email STARTS WITH 'ad' RETURN u`. The DB has a RANGE index on `User(email)`. Why is it slower than you expected?

`STARTS WITH` substring lookups are accelerated by a *TEXT* index, not a RANGE one — add a TEXT index on the column

Neo4j cannot answer prefix queries

Indexes and Constraints — Semantic Web Academy

Overview

Indexes and Constraints

Range, point, text indexes + UNIQUE / EXISTS constraints — your single biggest perf lever.

Why it matters

Without an index, the first MATCH in a query scans every node of the given label — fine for 10k nodes, catastrophic at 10M.

Going deeper

What each index type buys you and when to reach for it:

Index type	Best for	Don't use for
Range (BTREE default)	`=`, `<`, `>`, range scans on most types	Substring searches
Text	`STARTS WITH`, `CONTAINS`, `ENDS WITH`	Numeric ranges
Point	Geospatial distance and bounding-box	Non-geospatial properties
Token-lookup	`MATCH (n:Label)` with no property filter	Anything requiring property predicates
Composite	Queries that filter on N properties together	Single-property predicates (the composite isn't used)

A constraint also creates the matching index for free, so for any field you want unique you should prefer the constraint form over a bare CREATE INDEX.

Analogy

An index is the book's index, not the book's table of contents.

Without it, every lookup of 'where does Voltaire appear in this 800-page volume?' is a flip-every-page scan. With it, the lookup is one glance — the index says 'pages 12, 84, 309' and you jump straight to them.

A constraint is the publisher's promise that the index is honest: there will be only one entry for 'Voltaire' (UNIQUE), and every book chapter has a title (EXISTS). The DB enforces those promises so the index can stay correct.

Make it stick

Use the prompts below to anchor indexes and constraints to a real graph you own.

›List the top 3 read queries on your biggest label. For each, is the right index *type* in place?
›Where have you used `CREATE INDEX` when a `CREATE CONSTRAINT ... REQUIRE ... IS UNIQUE` would have given you both the index and an integrity check?
›Which composite index in your schema is *never* used because queries always filter on just one of its columns?

Reading in progress · 0 of 2 activities done