# Clustering

Running Memgraph in a clustered setup improves **resilience, fault-tolerance,
and uptime** for any graph-powered application. By deploying multiple Memgraph
instances, you can ensure continuous availability, even in the event of
failures.

Memgraph supports clustering through two main features:
- **Replication** (available in Memgraph Community)
- **High Availability (HA)** (available in Memgraph Enterprise)

Before configuring a cluster, we highly recommend reading the guide on **[how
replication works](https://memgraph.com/docs/clustering/replication/how-replication-works) in Memgraph**.
Understanding replication modes and choosing the right number of instances is
essential, as these decisions directly influence cluster performance and
reliability. In this guide you'll learn about the underlying implementation
and theoretical concepts behind Memgraph replication, including CAP
theorem, replication modes and synchronization mechanisms.

> **Note**
>
> Replication and high availability currently **work only in the [in-memory
> transactional storage
> mode](https://memgraph.com/docs/fundamentals/storage-memory-usage#in-memory-transactional-storage-mode-default)**.

## Replication <sup style={{ fontSize: '0.6em', color: '#888' }}>Community</sup>

Memgraph Community provides **[replication](https://memgraph.com/docs/clustering/replication) out of the
box**. A typical setup includes:
- **MAIN instance** – handles both reads and writes.
- **REPLICA instances** – read-only copies that can scale your query throughput.

You can run as many REPLICAS as needed. However, Community Edition **does not
provide automatic failover**, meaning that if the MAIN instance becomes
unavailable, manual intervention is required to promote a REPLICA.

To build a reliable architecture using Memgraph Community, you'll need to handle
failover logic yourself, either through custom tooling or orchestration
platforms.

In the [replication](https://memgraph.com/docs/clustering/replication) section of the docs you'll get
step-by-step instructions for setting up a replication cluster. Ideal for users
who want read scaling or multiple data copies. Note that high availability is
not included, so manual failover is required.

## High availability <sup style={{ fontSize: '0.6em', color: '#888' }}>Enterprise</sup>

Memgraph Enterprise includes everything from Community edition plus built-in
**[high availability](https://memgraph.com/docs/clustering/high-availability) management**. An Enterprise
cluster typically consists of:
- **MAIN instance**
- **REPLICA instances**
- **COORDINATOR instances** – powered by the Raft consensus protocol,
responsible for cluster state management and automatic leader election.

The COORDINATOR nodes ensure that the cluster can automatically recover from
failures, making true 24/7 uptime possible without manual failover procedures.

In the [high availability](https://memgraph.com/docs/clustering/high-availability) section of the docs
you'll learn how to configure, manage, and monitor a full HA cluster using
Memgraph Enterprise, including automatic failover and coordinator management.

## Concepts & internals

If you want to understand *how* clustering works under the hood before or while
you’re setting things up, these concept guides walk through the theory and
architecture behind Memgraph clustering:

- **Replication concepts**  
  Learn how Memgraph models replication from a distributed systems perspective:
  [CAP theorem](https://en.wikipedia.org/wiki/CAP_theorem), roles (**MAIN** /
  **REPLICA**), [replication
  modes](https://memgraph.com/docs/clustering/replication/how-replication-works#replication-modes)
  (**SYNC**, **ASYNC**, **STRICT_SYNC**), durability (snapshots, WALs, deltas),
  and how replicas recover and stay in sync.  
  → **[How replication works in
  Memgraph](https://memgraph.com/docs/clustering/replication/how-replication-works)**

- **High availability architecture** (Enterprise)  
  Deep dive into how Memgraph achieves automatic failover and 24/7 uptime using
  **coordinator instances**, the **Raft** consensus protocol, health checks,
  failover decision logic, and RPO/RTO considerations.  
  → **[How high availability
  works](https://memgraph.com/docs/clustering/high-availability/how-high-availability-works)**

- **Query routing in a HA cluster** (Enterprise)  
  Learn why plain `bolt://` isn’t enough in a clustered setup and how
  **Bolt+routing (`neo4j://`)** works with coordinators to always send writes to
  the current MAIN and transparently route reads across the cluster.  
  → **[Querying the cluster in high
  availability](https://memgraph.com/docs/clustering/high-availability/querying-the-cluster-in-high-availability)**

These pages focus on **theoretical and architectural aspects**. The sections
below focus on **how to actually configure and operate** your replication or
high availability cluster.

## FAQ

[Frequently asked questions](https://memgraph.com/docs/clustering/faq) about clustering, replication, and
high availability in Memgraph.