FAISS Vector DB

Overview

FAISS (Facebook AI Similarity Search) is a high-performance vector database service optimized for efficient similarity search and clustering of dense vectors. This service provides local, in-memory vector storage with advanced indexing algorithms for fast similarity searches, making it ideal for development, testing, and high-performance local applications.

Service Information

Property	Value
Service Name	Create FAISS vector DB
Status	Enabled
Compatible Nodes	Create knowledge DB, Real-time knowledge injector

When to Use This Service

FAISS Vector DB is ideal for:

Local development and testing of vector search applications
High-performance local search where speed is critical
Controlled infrastructure where you manage the deployment environment
Cost-sensitive applications avoiding cloud service fees
Offline operations without external service dependencies
Small to medium-scale vector databases that fit in available memory

Input Requirements

Embeddings: Pre-generated embedding vectors with consistent dimensions
Vector dimensions must be consistent across all embeddings in the database

Parameters

Index Configuration

Parameter	Type	Default	Options	Description
Index type	Choice	Inverted index	Inverted index, HNSW	Vector indexing algorithm for similarity search
Similarity metric	Choice	Cosine	Euclidean distance, Inner product, Cosine	Distance metric for vector comparisons
Quantization	Choice	None	Product quantization, None	Vector compression method to reduce memory usage