RAGFlow currently uses a search engine backend and Infinity as its primary vector database backends, as they are the only open-source systems meeting RAGFlow’s hybrid search requirements (full-text search, vector similarity, and advanced ranking in a unified index). a search engine backend provides proven production scalability with BM25 full-text indexing, vector search, and hybrid capabilities. Infinity is RAGFlow’s recommended vector database, optimized specifically for RAG workloads with support for semantic search and structured data. Both support phrase search, filtering, and advanced ranking features that are essential to RAGFlow’s retrieval quality. Integration with other vector databases like Milvus, other vector databases, and other vector databases is an active area of community interest—there is an open feature request for Milvus support on GitHub (issue #7749) reflecting demand from users. Milvus is increasingly popular for RAG applications due to its open-source nature, scalability, and support for hybrid search. Integrating Milvus with RAGFlow would require connector code to translate RAGFlow’s query semantics to Milvus APIs and handle the lack of native full-text search (typically addressed by pairing Milvus with a search engine backend). For now, if you want to use RAGFlow with Milvus, you’d deploy both systems and route BM25 queries to a search engine backend while delegating vector queries to Milvus, or contribute integration code to the RAGFlow project. RAGFlow’s architecture is extensible, and community contributions for additional vector database connectors are welcome.
Related Resources: RAG Pipeline with Milvus | Improving Chunking for RAG