Deploy Memgraph in the Google Cloud Platform (GCP)

This guide will show you how to deploy Memgraph on a Google virtual machine (VM). The guide will cover only the specific bits that are different from the general deployment guide you can find for native Linux or Docker deployments.

This guide assumes you have a Google Cloud Platform account and are familiar with the Google console. If you are not, you can follow the GCP Linux VM quick start guide documentation (opens in a new tab).

Creating a new GCP VM

The first step is to have a VM running on GCP, where you will deploy Memgraph. Since Memgraph works nicely with Linux distributions, it is recommended to use Linux-based VMs.

Below are some guidelines to consider if you are opening a new GCP VM for Memgraph.

Picking the OS

During the VM creation, you need to pick the OS you wish to have on the VM. If you are going to use the Memgraph native installation, you need to pick a Linux distribution that Memgraph supports.

Memgraph supports multiple Linux distributions. Memgraph packages for supported distributions and versions can be downloaded from the following page

Running Memgraph natively will bring some speed improvements compared to the Docker version of Memgraph.

If you are going to use the Memgraph Docker image, pick the Linux you are most familiar with.

Memgraph in Docker can be deployed both on x86 and ARM architecture. All native distributions work on x86 architecture. Some of the native distributions also work on ARM architecture (Debian, Ubuntu), some do not (Centos, Fedora). Check the direct download links for detailed information.

Picking the VM type

When creating the VM, you need to pick the VM type. If you run Memgraph in IN_MEMORY_TRANSACTIONAL (default mode) or in IN_MEMORY_ANALYTICAL storage mode, all data is stored in RAM. That means it would be good first to calculate how much memory you will need. There is also an easy-to-use calulator (opens in a new tab) for approximating memory usage. The good rule of thumb is for your server to have double the memory of what your storage was calculated to be. You can go with less if you do not have a demanding workload. A demanding workload would be a query that traverses half of the graph and returns half of the graph.

GCP has a collection of general purpose, memory-optimized, and compute-optimized instances. For more details on the instance types, you can check the GCP documentation (opens in a new tab).

The general-purpose instances are suitable for a wide range of workloads, and are good starting point for Memgraph deployment

The memory-optimized VMs are designed for large scale (> 1TB of RAM) in-memory databases, data analytics, and other memory-intensive applications. If you have a larger scale, consider using the memory-optimized instances since GCP offers them on TB scales. If your use-case requries lower memory configurations use general purpose instances, such as C4 family instances.

Instances vary based on the supported architectures, number of CPU cores, network bandwidth, block storage, etc. All hardware specs typically scale with the instance size. Memgraph is not demanding on the rest of the hardware specs as long as there is sufficent RAM.

If you are running Memgraph in ON_DISK_TRANSACTIONAL storage mode, you need to consider the instances optimized for storage (opens in a new tab).

Network Setup

When creating the VM, you need to set up the firewall for network access (opens in a new tab) and inbound port rules. By default, Memgraph uses port 7687 for the Bolt protocol. You need to open this port for TCP traffic and to allow connections to Memgraph.

If you change the default Bolt port, make sure to update the inbound port rules accordingly.

Also, if deploying Memgraph for Replication or High Availability, the ports for the replication and cluster management should also be open. In replication configuration each instance needs to have an open port 10000 for the replication. In a high availability configuration, each data instance needs to have an open port of 10000 for management and a port of 20000 for replication. Each coordinator instance*, needs to have open port 12000.

All instances need to have open port 7687 for the Bolt protocol.

Setup the storage

When creating the VM, you need to set up the disk storage. By default Memgraph stores all data to working RAM, but for the persistency between restarts Memgraph uses the disk storage to store snapshots, configurations, etc.

It is recommended to use the Persistent Disk storage (opens in a new tab). Persistent Disk volumes are durable, scalable, and can be attached to the VM. They will be persistent even if the VM is stopped and restarted.

Faster storage (SSDs) can lead to speedier snapshot creation and recovery times, which can be important on bigger scales (billion-sized graphs). Still, it is not critical for the Memgraph operating performance. Magnetic storage can also be used
on smaller scales.

The storage size depends on the amount of data you are going to store in your Memgraph instance and the number of active snapshots you want to keep alive. Memgraph will periodically create snapshots of the data and store them on disk. As the new snapshot is created, the oldest one is deleted. The number of snapshots you want to keep alive is configurable in the Memgraph configuration file.

The recommendation is to double the storage of the data you will store in Memgraph. If you are going to store 100GB of data in Memgraph, you should have at least 200GB of disk storage.

Installing Memgraph

Depending on the way you want to deploy Memgraph, native or via Docker, you need to follow the steps below:

docker run -p 7687:7687 memgraph/memgraph:latest

Manage Memgraph deployment

After Memgraph is installed and running on your GCP VM, Memgraph management in GCP VM is identical to the general guidelines that are described in the form of a native Linux Memgraph or Docker container Memgraph.

Depending on what you are using, you can follow the Linux or Docker deployment guide for more information on how to manage the Memgraph deployment.

Where to next?

Memgraph also supports deployment in the Kubernetes environment. If you are interested in deploying Memgraph in Kubernetes, you can follow the Memgraph Kubernetes installation guide.

To discuss GCP deployment and similar topics, join our Discord community (opens in a new tab).

Schedule a 30-min session with our engineers to discuss how Memgraph fits with your architecture. Our engineers are highly experienced in helping companies of all sizes to integrate and get the most out of Memgraph in their projects. Talk to us about data modeling, optimizing queries, defining infrastructure requirements or migrating from your existing graph database. No nonsense or sales pitch, just tech.