For AI agents: The documentation index is at https://docs.digitalocean.com/llms.txt. Markdown versions of pages use the same URL with index.html.md in place of the HTML page (for example, append index.html.md to the directory path instead of opening the HTML document).

Create a Vector Database Cluster

To create a vector database cluster:

1. Go to the Vector Databases page and click Create Vector Database.

2. In the Choose a Database engine section, select a database engine for your cluster:

   • Weaviate: A dedicated vector database. Use it when vector search is your primary workload, such as retrieval-augmented generation or semantic search. Weaviate is in public preview and uses the dedicated Vector Databases create flow described above.
   • OpenSearch: A search engine that supports hybrid vector and keyword search. Use it when you need both kinds of search in one query, or when you already run OpenSearch for logging or full-text search.
   • PostgreSQL: A relational database with the pgvector extension. Use it when your data is already in Postgres and vectors are a secondary dataset.

   To compare engines, see Choosing Between OpenSearch, Weaviate, and PostgreSQL with pgvector.

3. (PostgreSQL only) In the Choose an Edition section, select an edition for your cluster:

   • Standard Edition: Choose this option for development, testing, and simple production workloads with data sizes under 5 TiB.
   • Advanced Edition: Choose this option for production workloads that need higher availability, faster failover, scalability, and data sizes up to 30 TiB.

   Edition selection is permanent after you create the cluster.

4. In the Choose a database configuration section, select a configuration for your cluster. Configuration options vary by database engine.

   • For Weaviate, choose from the available Weaviate cluster configurations.
   • For OpenSearch, choose General Purpose - Dedicated CPU for production workloads that need consistent CPU performance, or Memory-Optimized - Dedicated CPU for larger vector datasets, higher-dimensional vectors, or high-query-volume workloads where the HNSW graph needs more memory.
   • For PostgreSQL, choose Basic - Shared CPU for development, testing, and smaller workloads; General Purpose - Dedicated CPU for production workloads that need consistent CPU performance; or Storage-Optimized - Dedicated CPU for larger datasets or workloads that need more storage capacity.

   If the configuration supports CPU options, select the CPU type for your cluster.

   If the configuration supports plan sizes, choose the plan size for your cluster.

   If the configuration supports storage size, select the amount of storage for the cluster. You can increase storage within the supported range without downtime.

   If the configuration supports storage autoscaling, enable storage autoscaling to automatically add storage before the cluster runs out of disk space. You can customize the threshold and storage increment.

   For PostgreSQL clusters that support standby nodes, use the Maximize uptime for critical workloads section to add standby nodes for high availability. Standby nodes provide automated failover and improve uptime for production workloads.

5. In the Choose a datacenter region section, choose a datacenter region for your cluster to live. For best performance, choose the same region as the application that sends queries to the cluster.

   Note

   The cluster uses the default VPC network for the selected datacenter region. Resources in the same VPC network can communicate securely over private IP addresses.

6. In the Finalize and create section, configure the cluster name and project:

   In the Choose a unique database cluster name field, use the generated name or enter a unique name for the cluster. Names must be 3-63 characters, lowercase, and contain only letters, numbers, and dashes.

   From the Select a project dropdown, choose the project for the cluster.

7. In the Total monthly cost section, review the estimated monthly and hourly cost for the cluster, including the primary node, storage size, autoscale increment, and autoscale threshold.

8. Click Create Database Cluster.

Provisioning can take several minutes depending on the engine and cluster size.

Secure the Cluster

OpenSearch and PostgreSQL vector database clusters reject connections until you add at least one trusted source. A trusted source is an IP address, CIDR range, or DigitalOcean resource that is allowed to connect to the cluster.

Weaviate clusters use a public HTTPS/gRPC endpoint, TLS on port 443, and a Weaviate API token sent as a bearer token with each request.

To add a trusted source to an OpenSearch or PostgreSQL cluster:

1. Go to the Vector Databases page and select the database you want to add a trusted source to.

2. On the cluster’s Overview page, click the Network Access tab, and then click Add Trusted Sources to open the Add Trusted Sources window.

3. In Add Trusted Sources, choose how to grant access to the cluster:

   • Enter specific IP addresses or CIDR notations: Enter specific IP addresses or a CIDR range. Or click My current IP address to use the Quick Add option, which adds your machine’s current IP address.
   • Quick select Droplets, Kubernetes clusters, Apps, and tags: Use the search to find a resource or click the dropdown menu and select a resource from the list.

   We recommend adding a description for each trusted source so you can identify it later.

4. Click Add Trusted Sources.

After you add a trusted source, applications from that source can connect to the cluster.

Run a Query

Before you run queries, collect the cluster connection details:

1. Go to the Vector Databases page and select the cluster you want to connect to.

2. On the cluster’s Overview page, in the Connection Details section, copy the cluster’s username, password, host, and port.

3. Click the Network Access tab, and then copy and save the trusted source you want to use to connect to the cluster.

After you collect the connection details, follow the engine-specific guide to connect to the database, create a vector index, insert embeddings, and run a vector query:

• OpenSearch Vector Search
• PostgreSQL with pgvector
• Managed Weaviate

Destroy a Cluster

DigitalOcean Vector Databases are billed by the hour. Destroy clusters when you no longer need them to avoid ongoing charges. Destroying a cluster permanently deletes all indexes and vectors.

To destroy a cluster:

1. Go to the Vector Databases page and locate the cluster you want to destroy.

2. To the right of the cluster, click …, and then click Destroy.

3. In the Destroy vector database window, enter the name of the cluster to confirm deletion.

4. In the Tell us why you’re destroying this database cluster section, choose an option.

5. Click Destroy.