logo

Documentation

Vector Database Connectors

Vector Database Connectors provide powerful integration capabilities for various vector databases and similarity search engines. These connectors support multiple vector database solutions including Milvus, Qdrant, Astra DB, Weaviate, Pinecone, and Chroma DB, enabling efficient vector similarity search and retrieval for your RAG applications.

1.1 Milvus

Milvus Configuration Interface

Milvus Connector Interface

Description

The Milvus connector enables seamless integration with Milvus vector database, providing high-performance similarity search capabilities. This connector supports collection management, vector indexing, and efficient querying of high-dimensional vector data.

Useful Links

Use Cases

  • Similarity search in large-scale vector datasets
  • Real-time vector search applications
  • Image and video similarity search
  • Semantic text search and recommendation systems
  • Feature vector matching and retrieval

Inputs

  • Collection Name: Name of the Milvus collection (required)

    Example: image_vectors

  • Collection Description: Description of the collection (optional)

    Example: Image feature vectors for similarity search

  • Connection URI: Milvus server connection URI (required)

    Example: http://localhost:19530

  • Token: Authentication token (optional)

    Example: milvus_token_123

  • Other Connection Arguments: Additional connection parameters (optional)

    Example: timeout: 30, retry_interval: 1000

  • Primary Field Name: Primary key field name (required)

    Example: id

  • Text Field Name: Text field name (optional)

    Example: text

  • Vector Field Name: Vector field name (required)

    Example: vector

Implementation Notes

  • Configure appropriate index types for your use case
  • Optimize collection schema for performance
  • Implement proper error handling and retries
  • Monitor collection metrics and performance
  • Use batch operations for better throughput

1.2 Qdrant

Qdrant Configuration Interface

Qdrant Connector Interface

Description

The Qdrant connector provides integration with Qdrant vector database, offering advanced vector similarity search capabilities with filtering support. This connector enables efficient management and querying of vector collections with metadata filtering.

Useful Links

Inputs

  • Host: Qdrant server host (required)

    Example: localhost

  • Port: Server port number (required)

    Example: 6333

  • gRPC Port: gRPC port number (optional)

    Example: 6334

  • API Key: Authentication API key (optional)

    Example: qdrant_api_key_123

  • Prefix: Collection name prefix (optional)

    Example: prod_

1.3 Astra DB

Astra DB Configuration Interface

Astra DB Connector Interface

Description

The Astra DB connector enables integration with DataStax Astra DB, providing scalable vector search capabilities with the reliability of Apache Cassandra.

Useful Links

Inputs

  • Astra DB Application Token: Authentication token (required)

    Example: AstraCS:token123

  • API Endpoint: Astra DB API endpoint (required)

    Example: https://api.astra.datastax.com

  • Collection: Vector collection name (required)

    Example: vector_collection

1.4 Weaviate

Weaviate Configuration Interface

Weaviate Connector Interface

Inputs

  • Weaviate URL: Server URL (required)

    Example: http://localhost:8080

  • API Key: Authentication key (optional)

    Example: weaviate_api_key_123

  • Index Name: Class name in Weaviate (required)

    Example: Article

  • Text Key: Text field key name (required)

    Example: content

1.5 Pinecone

Pinecone Configuration Interface

Pinecone Connector Interface

Inputs

  • Index Name: Pinecone index name (required)

    Example: my-index

  • Namespace: Index namespace (optional)

    Example: production

  • Pinecone API Key: Authentication key (required)

    Example: pinecone_api_key_123

1.6 Chroma DB

Chroma DB Configuration Interface

Chroma DB Connector Interface

Inputs

  • Collection Name: Name of the collection (required)

    Example: embeddings

  • Persist Directory: Directory for persistence (required)

    Example: /data/chroma

  • Search Query: Query for similarity search (required)

    Example: "semantic search query"

  • Ingest Data: Data to be ingested (optional)

    Example: text content, metadata:source:article, author:John Doe

1.7 FAISS

FAISS Configuration Interface

FAISS Connector Interface

Inputs

  • Index Name: Name of the FAISS index (required)

    Example: semantic_index

  • Persist Directory: Storage location for index (required)

    Example: /path/to/faiss/index

  • Allow Dangerous Deserialization: Safety flag (optional)

    Example: false

1.8 Supabase

Supabase Configuration Interface

Supabase Connector Interface

Inputs

  • Supabase URL: Supabase instance URL (required)

    Example: https://your-project.supabase.co

  • Supabase Service Key: Service role API key (required)

    Example: supabase_service_key_123

  • Table Name: Vector table name (required)

    Example: embeddings

  • Query Name: Query identifier (optional)

    Example: similarity_search

Important Implementation Notes:

  • Choose the appropriate vector database based on your scaling needs and use case
  • Implement proper security measures including API key management
  • Monitor performance metrics and implement appropriate indexing strategies
  • Use batch operations when possible for better performance
  • Implement proper error handling and retry mechanisms
  • Consider data backup and disaster recovery procedures