Neo4j · Rate Limits

Neo4J Rate Limits

Neo4j is a graph database; capacity scales with instance size (RAM, CPUs, storage) rather than per-second HTTP throttling. The Aura management API and Query API do not publish explicit RPS limits. The AuraDB Free tier limits node and relationship counts. Bolt protocol concurrency depends on the chosen instance size.

3 Limits Throttle: 429
Rate LimitingGraph DatabaseAuraDB

Limits

AuraDB Free node + relationship count instance
nodes_relationships
see AuraDB Free tier docs
Free tier limits node and relationship counts rather than request rate.
AuraDB Professional / Business Critical capacity instance
varies
bound by selected GB RAM, CPU, storage
Throughput scales with selected size (1-512 GB RAM, 1-96 CPU cores).
Aura Console / Query API account
varies
see Aura API specification
No explicit RPS limit published in the public Aura API specification.

Policies

Capacity-based scaling
Right-size AuraDB instance memory and CPU to handle target query throughput; Neo4j sizing guidance recommends measuring p95 query latency under load and scaling vertically.
Connection pooling
Use the official Neo4j drivers' connection pool; do not open a new Bolt session per request. Default pool size is 100 connections per driver instance.
Query timeouts
Set `dbms.transaction.timeout` to bound long-running Cypher queries on self-managed Enterprise; AuraDB applies platform timeouts.
Backpressure on free tier
Free tier reaches node/relationship caps; queries that would exceed the cap fail. Upgrade to Professional for production volumes.

Sources