Elasticsearch
A distributed search and analytics engine that now includes vector search capabilities. Originally built for full-text search, Elasticsearch added dense vector support and kNN search, making it a hybrid search platform combining traditional text search with semantic vector search.
Implements
Concepts this tool claims to implement:
- Hybrid Search primary
Combines BM25 full-text search with dense vector kNN search. Reciprocal Rank Fusion (RRF) for merging text and vector results. Native support for combining keyword and semantic search.
- Vector Search primary
dense_vector field type with kNN search. Supports HNSW algorithm for approximate nearest neighbor. Cosine, dot product, and L2 similarity metrics.
- Retriever secondary
Used as retrieval backend for RAG systems. LangChain, LlamaIndex, and other frameworks have Elasticsearch integrations for retrieval.
Integration Surfaces
Details
- Vendor
- Elastic
- License
- SSPL / Elastic License 2.0
- Runs On
- local, cloud, hybrid
- Used By
- human, agent, system
Links
Notes
Elasticsearch's vector capabilities were added later but are now production-ready. Main advantage is combining existing text search expertise with vector search. Elastic Cloud provides managed hosting. License changed from Apache 2.0 to dual SSPL/Elastic in 2021.