Running Neo4j?

Neo4j cluster running hot or queries timing out? We sort it out.

Graph databases shine on the queries that are miserable in SQL and punish you on the queries that are easy. Teams come to us when a Cypher query that worked at ten thousand nodes falls over at ten million, when the heap keeps crashing under a mixed workload, or when a causal cluster refuses to elect a leader. We get it stable and then we make it quick.

  • Since 2015Fixing technology
  • FreeEstimates
  • US teamReal engineers

neo4j-debug.log

still failing

What it is

Neo4j at a glance

Neo4j is a native graph database built around the property graph model: nodes, relationships, and properties on both. It was founded in Sweden in 2007 by Emil Eifrem and others, and its query language Cypher became the basis of the ISO GQL standard ratified in 2024.

The database runs as a single instance or as a causal cluster with raft based consensus, and offers a managed service called AuraDB on AWS, GCP and Azure. It is used where traversal queries dominate: fraud rings, recommendation graphs, identity resolution, access control, knowledge graphs feeding retrieval augmented generation. For pure document or tabular workloads it is usually the wrong tool, which is a conversation we have often.

Type
Native graph database with Cypher query language
First released
2007, version 1.0 in 2010
License
GPLv3 Community edition, commercial Enterprise, managed AuraDB
Maintained by
Neo4j, Inc.
Typically used for
Fraud detection, recommendations, identity, knowledge graphs, network analysis

The honest version

Neo4j pros and cons

No technology is all upside. This is where Neo4j earns its place and where it tends to cause trouble.

Where Neo4j is strong

  • Traversals are genuinely fastMulti hop queries that would be a nightmare of joins in Postgres run in milliseconds against a graph that is indexed and warm, which is the whole reason for using it.
  • Cypher is readablePattern matching with ASCII art arrows reads closer to what the query is actually doing than nested joins, which makes code review and debugging easier.
  • ACID transactions on a graphUnlike several other graph stores, Neo4j commits atomically with isolation levels you can reason about, which matters for anything touching money or identity.
  • Good toolingNeo4j Browser, Bloom and the data importer let non engineers explore the graph without writing code, which keeps the model close to the business.
  • Managed AuraDB availableIf you want the graph without the Java tuning, AuraDB takes heap and cluster operations off your plate for a per hour price.
  • Growing GraphRAG use casePairing a knowledge graph with an LLM for retrieval is a credible modern pattern, and Neo4j is the most common store teams reach for.

Where Neo4j bites

  • It is a JVM databaseHeap tuning, garbage collection pauses and page cache sizing are part of running it. If your team is Node and Postgres by default, this is a new operational skill set.
  • Queries can get slow in unobvious waysA missing index, a cartesian product in a pattern, or a planner that picked the wrong starting label can take a query from twenty milliseconds to twenty seconds without any error at all.
  • Cluster operations need careCausal cluster failover, store copy and rolling upgrades work, but each needs a deliberate runbook. A rushed restart can leave a follower with a corrupted transaction log.
  • Enterprise licensing is not cheapClustering, fine grained security, hot backups and more of the advanced features are Enterprise only. Pricing scales with cores, which means the bill goes up when you do.
  • Ecosystem is narrower than SQLORMs, migration tools, observability and reporting integrations exist but are thinner than the Postgres or MySQL equivalents. Expect to do some gluing yourself.
  • Hiring is a smaller poolEngineers comfortable with Cypher, graph modelling and JVM tuning together are rarer than SQL generalists, and the skill set takes months rather than weeks to build.

Sound familiar?

Common Neo4j problems we fix

The Neo4j problems that bring people to us, and what you have once each one is fixed.

OutOfMemoryError: Java heap space

The heap keeps exhausting

Large result sets materialised in a single transaction, a query returning all shortest paths across the graph, or a heap sized for a dev laptop running production load.

Heap and page cache sized for the workload

TransactionTimedOut

Cypher queries time out under load

A query that works on cold data falls over on peak traffic because the planner chose a full label scan or a cartesian product that nobody caught in review.

Plans inspected, indexes in place

DeadlockDetected

Writes deadlock during spikes

Concurrent transactions acquiring locks on the same relationships in opposite orders, which Neo4j correctly aborts but which the application then retries badly.

Lock order fixed, retries sensible

Neo.TransientError.Cluster.NotALeader

A causal cluster will not elect a leader

Network partitions, clock skew between members, or a follower that lost its store and cannot catch up through copy. The cluster is up but will not accept writes.

Cluster healthy, leader stable

CorruptedLogFileException

Transaction log corruption after an unclean shutdown

A kernel panic, a full disk or a mistimed power event left the transaction log in a state the recovery path cannot finish, and the database will not start.

Recovered from logs or restored

cartesian product between disconnected patterns

The planner generates a cartesian product

Two patterns in the same MATCH clause that share no variable, so every node on one side pairs with every node on the other, which quickly becomes billions of rows.

Pattern rewritten, planner guided

Index is in populating state

New indexes do not seem to help

An index was created but the application started querying before it finished populating, or the schema has an index on label and property that the query does not touch.

Indexes verified, queries routed through them

page cache hit ratio: 0.43

Hot reads keep going to disk

The page cache is sized far smaller than the working set, so every query fetches from storage, and performance looks nothing like a benchmark on the same hardware.

Cache tuned to the working set

What we do

Neo4j help, from rescue to tuning

Fix one stuck query, or hand us the whole cluster and its ops.

Rescue
A cluster that will not accept writes, a heap that keeps crashing, or a query that stopped returning. We stabilise it first and write up what was going on afterward.
Query and schema tuning
PROFILE and EXPLAIN on the queries that matter, indexes created where they help, patterns rewritten where they do not.
Graph modelling
Working out which concepts are nodes, which are relationships, and where to denormalise, so the data layout matches how you actually query it.
Cluster operations
Causal cluster setup, rolling upgrades, hot backups and recovery drills, documented so the next outage is a runbook rather than a rescue.
GraphRAG integrations
Wiring a knowledge graph into a retrieval augmented generation pipeline, with embeddings, traversals and LLM prompts that actually use the graph structure.
Managed Tech
Backups, upgrades, heap and query monitoring, and an engineer on call who knows your graph when something starts behaving oddly.

How it works

From first message to fixed

  1. Look Free estimate

    Tell us what's going wrong. We look at it, then tell you plainly what's wrong and what it will cost.

  2. Fix

    We fix the cause, not the symptom. If something is down, getting it back up comes first.

  3. Harden

    Backups, monitoring, updates and access, so the same problem can't quietly come back.

  4. Hand back or stay on

    You get a plain-language write-up. Or we stay on for hosting, upkeep and new work.

Often found together

Running Neo4j alongside something else?

We work on the whole stack, so the problem can be wherever it actually is.

Questions

Neo4j questions

Something else? Call+1 737 727 7327

Do we actually need a graph database?

Only if your queries are genuinely about relationships. Friend of friend, shortest path, fraud ring detection and recommendation graphs are good fits. Tabular reporting, time series and simple document lookups are almost always cheaper and faster to run in Postgres. We will say so if your use case is in the second bucket.

Why are our Cypher queries slow?

Three usual suspects. A missing index on the label and property the query starts from, a cartesian product hidden inside a seemingly innocent MATCH, and a planner that picked a scan where a seek would have worked. PROFILE shows which of the three it is, and the fix is usually small once the cause is clear.

Can you take over a Neo4j cluster we inherited?

Yes, that is common work. We inventory the deployment, work out what the real backup and recovery story is, bring heap and page cache onto sensible values, and document the cluster so your team is no longer dependent on whoever set it up.

Should we self host or use AuraDB?

AuraDB saves you the JVM tuning, upgrade cycles and cluster operations, and pays back quickly for teams without a database specialist on staff. Self hosted makes sense when you need custom configuration, strict data residency or you already run a lot of JVM workloads. We will cost both against your traffic rather than guess.

How do we use Neo4j for GraphRAG with an LLM?

The graph holds the entities and relationships, the LLM handles language. We model the domain as a graph, generate embeddings on nodes where semantic similarity helps, and wire Cypher traversals into the retrieval step so the model sees connected context rather than isolated chunks. The result is answers grounded in your structure rather than a vector search in the dark.

How much does Neo4j help cost?

Estimates do not cost anything. Send a query plan, a heap dump or just a description of what the cluster is doing, and we come back with a price up front.

Free estimate

Neo4j giving you trouble?

Tell us what you're running and what's going wrong. A person reads it and comes back with an estimate.

  • No charge to find out what it will take
  • If it's urgent, say so

Replies come from a person. By sending, you authorize us to email you back. No marketing.