Skip to main content


LICENSEΒ Weaviate on Stackoverflow badgeΒ Weaviate issues on Github badgeΒ Weaviate version badgeΒ Weaviate total Docker pulls badgeΒ Go Report Card

There are three ways you can run Weaviate, and they are -we believe- pretty straightforward.

  1. Weaviate Cloud Service
  2. Docker
  3. Kubernetes

Weaviate Cloud Service​

The fastest way πŸš€ to create a new Weaviate instance – from scratch – is to use the Weaviate Cloud Service (aka, the WCS). The WCS is a completely managed service, so you don't have to install or maintain anything to run Weaviate. Currently, the service is in private beta, but (🀫) if you log in to the Weaviate Cloud Console, you can create a free sandbox to play around with.

If you are itching to get started with WCS, just skip to the WCS hands-on section.

Running Weaviate yourself​

Alternatively, if you prefer to install and deploy Weaviate yourself, then you can work with either Docker or Kubernetes.


Working with Docker is great if you are building an application around Weaviate and want to run Weaviate on your local machine or in the cloud. If you have Docker already installed, you could have it all up and running in seconds (minutes if you use a prepackaged transformers module).

We even have a handy step-by-step configurator, which lets you pick your configuration, and as a result you will receive a command to spin up your docker setup.


Generally, we recommend using Kubernetes to deploy Weaviate for any long-running deployments or those with specific availability expectations, such as production use cases.

For local development or personal evaluation, using Docker Compose will most likely be sufficient.

Using Weaviate with Kubernetes is out of scope for this tutorial. You can find more info on Kubernetes here.

Self-deployment instructions​

The installation and configuration with Docker and Kubernetes is out of scope for this tutorial, as all code examples include references to a WCS instance. To learn about deploying Weaviate locally, follow the installation documentation page.

If you feel comfortable making a few changes to the code examples in the tutorial and you prefer to run Weaviate locally, you can set up Weaviate with Docker in two steps:

  1. Get docker-compose.yml configuration file by calling:
    curl -o docker-compose.yml ""
  2. Spin up docker
    docker-compose up -d

Then you can continue with the tutorial and skip to the Schema page.

WCS hands-on​

To create a new Weaviate instance on the Weaviate Cloud Service, we need to follow these steps:

  1. Sign in to WCS
  2. Create a Weaviate Cluster
  3. Test the connection

Sign in to WCS​

In order to access WCS, navigate to the Weaviate Cloud Console and "Sign in with the Weaviate Cloud Service", where you will be able to create and manage your Weaviate Clusters.

No account, no problem​

If you don't have an account with WCS yet, click the "Don't have an account? Sign Up" link and create an account.

Register link

Create a Weaviate Cluster​

To create a new Weaviate Cluster:

  1. Press the "Create a Weaviate Cluster" button

  2. Configure the cluster:

    1. Set the name for your cluster – note: The name will become part of the URL we will use to access this instance. Please use a different name than "getting-started".
    2. Leave the Subscription Tier as Sandbox - note: The sandbox is free, but it will expire after 30 days
    3. Leave the Weaviate Version as the latest
    4. Leave the Standalone Or Modules as Standalone, no Modules
    5. change the OIDC Authentication to Disabled

    Cluster Configuration

  3. Finally, press Create.

This will start the process to create a new cluster. The status indicator on the left will show the progress (in %); after a few minutes, you should see a green tick βœ”οΈ - indicating that the cluster is ready.

Test the connection​

To test the connection, click on the Cluster Id link, which will open a new page in your browser and display all the available endpoints. Weaviate Cluster


For every endpoint, there is a documentationHref link, which points us to relevant documentation pages.

Now, we are ready to start working with Weaviate. πŸŽ‰

Running Weaviate yourself​

When running Weaviate yourself in production, you want to make sure you select the right hardware to run it on.Β  The benchmark pages in the documentation are helpful for this (more about this in this guide) too, take the following things into account when choosing the right hardware:

  1. Disks – use SSD disks if possible. Weaviate works more efficiently onΒ solid state disks than on spinning disks. 0. SSD disks come in a wide variety of types and price ranges. You might want to experiment with this, but based on our experience, there is a marginal return when spending large amounts of money onΒ extreme SSD types. 0.Avoid network storage and go for block storage. Internally we use; 0. gp3 on Amazon Web Services 0. premium-rwo for Google Cloud Platform 0. Premium SSD for Microsoft Azure
  2. Memory – make sure there is enough available to store the indices. To calculate to amount of memory needed for your vectors. Follow this calculation. You can learn more about memory usage in Weaviate here.
  3. CPUs – adding more CPUs increases import speed or query time. Setting up monitoring for your Weaviate instance will help you determine if you need more or fewer CPUs in your setup.



If you're new to Weaviate but familiar with Kubernetes. It might be an idea to use the Docker-compose configurator first to see how Weaviate is structured.

For this one, you need to understand how Kubernetes works; these are just two handy things to know.

  1. If you want to use Weaviate in combination with modules, it might be handy to check out the Docker guide first. It will align with the Helm charts.
  2. You find all detailed Kubernetes instructions here.
version: '3.4'
image: semitechnologies/weaviate:1.14.0
- 8080:8080
restart: on-failure:0
CONTEXTIONARY_URL: contextionary:9999
PERSISTENCE_DATA_PATH: '/var/lib/weaviate'
DEFAULT_VECTORIZER_MODULE: 'text2vec-contextionary'
ENABLE_MODULES: text2vec-contextionary
EXTENSIONS_STORAGE_ORIGIN: http://weaviate:8080
image: semitechnologies/contextionary:en0.16.0-v1.0.2

Working with GPU-enabled containers​

Let's just cut straight to the chase; running modules with ML models yourself (i.e., a Weaviate module where the model is encapsulated inside the module) on a CPU is just not going to work well. It's sloooow 🐌.

You can use the Kubernetes set up with modules or run Weaviate with Docker on a machine with a GPU (this Github Gist contains an installation script to install Docker Compose with GPU support on a Debian machine)


  • You can run Weaviate with Weaviate Cloud Service, Docker or Kubernetes.
  • Weaviate Cloud Service is a managed Weaviate SaaS - great for development and production.
  • Docker alows you to set up Weaviate on your machine or in the cloud – great for development and production.
  • Kubernetes is great for long-running Weaviate deployments
  • You have a working instance of Weaviate in WCS.

More Resources​

If you can't find the answer to your question here, please look at the:

  1. Frequently Asked Questions. Or,
  2. Knowledge base of old issues. Or,
  3. For questions: Stackoverflow. Or,
  4. For issues: Github. Or,
  5. Ask your question in the Slack channel: Slack.