Welcome to Quiver! 🏹¶
-
Blazing Fast
Search through millions of vectors in microseconds -
Rich Metadata
Combine vector search with powerful SQL filtering -
Production Ready
Built for reliability, scalability, and performance -
Secure & Robust
Encryption, backups, and comprehensive monitoring
What is Quiver?¶
Quiver is a high-performance vector database built in Go. It's designed to make similarity search on high-dimensional vectors not just fast, but ridiculously fast. Whether you're building a recommendation system, semantic search, or AI-powered application, Quiver gives you the tools to find what you're looking for in microseconds.
// Initialize Quiver
idx, _ := quiver.New(quiver.Config{
Dimension: 128,
StoragePath: "data.db",
}, logger)
// Add vectors with metadata
idx.Add(1, vector, map[string]interface{}{
"category": "science",
"name": "black hole",
})
// Search for similar vectors
results, _ := idx.Search(queryVector, 10)
Why Quiver?¶
Speed That Makes You Go "Whoa!" 🤯¶
Quiver is built on the HNSW algorithm, one of the fastest approximate nearest neighbor search methods available. But we didn't stop there - we've optimized every aspect of the system to deliver search results in microseconds, even with millions of vectors.
Not Just Vectors, But Context 🧩¶
Unlike many vector databases that only store vectors, Quiver integrates with DuckDB to provide rich metadata storage and querying. This means you can combine vector similarity with traditional SQL filtering for powerful hybrid searches.
// Find similar vectors that match specific metadata criteria
results, _ := idx.SearchWithFilter(queryVector, 10,
"category = 'science' AND tags LIKE '%physics%'")
Built for the Real World 🌍¶
Quiver isn't just a research project - it's built for production use. With features like:
- Automatic persistence and recovery
- Scheduled backups with rotation
- Encryption for data at rest
- Comprehensive monitoring and metrics
- HTTP API for easy integration
Getting Started¶
Ready to give Quiver a shot? Check out our Quick Start Guide or dive into the Installation Instructions.
Performance¶
Quiver is fast. Here are some benchmarks on an M2 Pro CPU:
Operation | Throughput | Latency | Memory/Op |
---|---|---|---|
Search | 16.9K ops/sec | 59μs | 24.2 KB |
Add | 6.4K ops/sec | 156μs | 20.9 KB |
Hybrid Search | 4.8K ops/sec | 208μs | 80.6 KB |
Community & Support¶
- GitHub Issues - Bug reports and feature requests
- GitHub Discussions - Questions and community discussions
- Contributing Guide - How to contribute to Quiver