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).

Getting Started

Quickstart guides for using DigitalOcean Managed OpenSearch as a vector database.

How-Tos

How to create OpenSearch vector clusters, build k-NN indexes, run hybrid searches, and register remote embedding models.

Concepts

Background reading on vector embeddings, HNSW, engines, hybrid search, and where embeddings come from, specific to DigitalOcean Managed OpenSearch.

DigitalOcean Managed OpenSearch is the same engine available under Managed Databases, entered through the DigitalOcean Vector Databases flow. OpenSearch 2.19 bundles the k-NN, ML Commons, and Neural Search plugins, so you can create a k-NN index, register a remote embedding model, or run hybrid (vector plus keyword) search as soon as provisioning completes.

When to Choose OpenSearch for Vector Workloads

OpenSearch is the recommended engine when you need:

• Hybrid search: BM25 keyword scoring combined with vector similarity, with per-query weights and score normalization.
• Server-side embeddings: ML Commons connectors to OpenAI, Bedrock, Cohere, SageMaker, and custom HTTP endpoints, so your application never has to embed text itself.
• Full-text search alongside vectors: OpenSearch is a general-purpose search engine first; vectors are a first-class field type on any index.
• Log, metric, or trace co-location: If you already run OpenSearch for observability, adding a k-NN index reuses the same cluster.

For pure vector-native workloads without keyword search, consider Weaviate. For small vector datasets co-located with relational data, consider PostgreSQL with pgvector.

What Is Included by Default

Reusing Existing OpenSearch Clusters

OpenSearch clusters created under either Vector Databases or Managed Databases are functionally identical. If you already operate a Managed OpenSearch cluster, you can turn it into a vector database by enabling k-NN on an index. No cluster recreation required.

Latest Updates

27 April 2026

• DigitalOcean Vector Databases is now generally available in Data Services that groups managed engines for vector similarity search. The launch includes:

  • Weaviate in private preview for retrieval-augmented generation and semantic search workloads. Available to opted-in customers; preview clusters are not billed.
  • OpenSearch with the k-NN, ML Commons, and Neural Search plugins for hybrid (vector plus keyword) search and remote embedding models. Uses the existing Managed Databases OpenSearch engine.
  • PostgreSQL with the vector (pgvector) and vectorscale (pgvectorscale) extensions for vector similarity search alongside relational data.

  For an overview and guidance on choosing an engine, see Vector Databases.

17 September 2025

• Now in public preview, you can now enable storage autoscaling on all Managed Database engines. To enable autoscaling, see our resizing guides for MySQL, PostgreSQL, MongoDB, OpenSearch, and Kafka.

• Storage autoscaling is now in general availability. Additionally, you can now reduce your cluster’s amount of additional storage, as long as the new storage size is greater than or equal to the latest backup’s size plus any data growth since then and a 25% buffer.

28 February 2025

• All managed databases except MongoDB now support up to 2,000 IP addresses as trusted sources. To add trusted sources, see our guides for MySQL, PostgreSQL, Caching, MongoDB, OpenSearch, and Kafka.

For more information, see the full release notes.