Deploying Weaviate
This is a new section being created, please bear with us as we add in new content!
Weaviate is available as a hosted service, Weaviate Cloud (WCD), or as a self managed instance. If you manage your own instance, you can host it locally or with a cloud provider. Self-managed instances use the same Weaviate Database as WCD.
If you are upgrading from a previous version of Weaviate, see the Migration Guide for any changes that may affect your installation.
Weaviate offers multiple deployment options to satisfy your specific use case in production.
This section hosts common deployment topics, including Kubernetes, cloud providers, and best practices, along with detailed tutorials and how-to guides. Weaviate is designed for:
- Scalability – Handle billions of vector data points efficiently.
- High-Performance Search – Power AI applications with real-time vector retrieval.
- Flexible Integrations – Connect with various machine learning models and data sources.
- Cloud & On-Prem Deployment – Deploy on Weaviate Cloud, Kubernetes, or managed cloud services.
Deployment Options
Choose the best deployment method based on your needs:
- From evaluation (sandbox) to production
- Serverless (infrastructure managed by Weaviate)
- (Optional) Data replication (high-availability)
- (Optional) Zero-downtime updates
- For local evaluation & development
- Local inference containers
- Multi-modal models
- Customizable configurations
- From development to production
- Local inference containers
- Multi-modal models
- Customizable configurations
- Self-deploy or Marketplace deployment
- (Optional) Zero-downtime updates
- From evaluation (sandbox) to production
- Serverless (bill through AWS)
- Kubernetes (billed through AWS)
- Self-hosted on EKS
- From evaluation (sandbox) to production
- Serverless (bill through GCP)
- Kubernetes (billed through GCP)
Methods
In addition to the above deployment methods, we also have the following:
- Weaviate Cloud: Managed services for development and production environments.
- Snowpark Container Services Deploy Weaviate in Snowflake's Snowpark environment.
- Embedded Weaviate: Experimental. Embedded Weaviate is a client based tool.
Configuration files
Docker Compose and Kubernetes use yaml files to configure Weaviate instances. Docker uses the docker-compose.yml file. Kubernetes relies on Helm charts and the values.yaml file. The Weaviate documentation also calls these files configuration yaml files.
If you are self-hosting, consider experimenting on a small scale with Docker and then transferring your configuration to Kubernetes Helm charts when you are more familiar with Weaviate.
Versions
Weaviate version availability may differ across deployment options, Weaviate Cloud generally has the latest versions of all deployment methods.
DISCLAIMER: Release candidate images and other unreleased software are not supported.
Unreleased software and images may contain bugs. APIs may change. Features under development may be withdrawn or modified. Do not use unreleased software in production.
To run an unreleased version of Weaviate, edit your configuration file to use the unreleased image instead of a generally available image. The GitHub releases page lists generally available and release candidate builds.
For example, to run a Docker image for a release candidate, edit your docker-config.yaml to import the release candidate image.
image: cr.weaviate.io/semitechnologies/weaviate:1.23.0-rc.1
When you try upcoming features, please provide feedback. Your comments are appreciated and help us to make Weaviate more useful for you.
Related pages
Questions and feedback
If you have any questions or feedback, let us know in the user forum.