vearch

Vearch is a cloud-native distributed vector database designed for efficient similarity search of embedding vectors in AI applications.

• Website/Source: https://github.com/vearch/vearch
• Category: vector-database-engines
• Tags: open-source, distributed, vector-search, ai
• License: Apache-2.0

Features

• Distributed & Cloud-Native: Built for scalable, distributed deployments.
• Hybrid Search: Supports both vector search and scalar filtering.
• Performance: Fast vector retrieval, capable of searching millions of objects in milliseconds.
• Scalability & Reliability: Supports replication and elastic scaling out.
• Replication: Raft-based partition replication for reliability.
• RESTful APIs: Provides APIs for upsert, delete, search, and query operations.
• Multiple SDKs: Official SDKs for Python and Go; Java SDK under development.
• Integration with AI Frameworks:
  • Langchain (Python)
  • LlamaIndex
  • Langchaingo (Go)
  • LangChain4j (Java)
• Visual Search Support: Can be used to build large-scale visual search systems (e.g., indexing billions of images).
• Core Engine: Uses Gamma, based on Faiss, for vector storage, indexing, and retrieval.
• Easy Deployment:
  • Kubernetes (Helm charts)
  • Docker Compose (standalone and cluster modes)
  • Docker and source compilation supported
• Component Architecture:
  • Master: Schema management, metadata, resource coordination
  • Router: REST API, request routing, result merging
  • PartitionServer (PS): Hosts document partitions with replication
• OpenAPI Support: For API documentation and integration
• Backup & Compression: Supports backup with zstd compression
• Flamegraph UI: Web UI support for performance flamegraphs

Pricing

Vearch is open-source and free to use under the Apache-2.0 license. No paid plans are mentioned.

Usage Examples

• As a memory backend for AI and RAG (retrieval-augmented generation) applications
• As a vector store for frameworks like Langchain and LlamaIndex
• Building large-scale image and visual search systems

Deployment

• Kubernetes: via Helm repository or local charts
• Docker Compose: standalone and cluster modes
• Docker and source compilation options

References

• Vearch Documentation
• API Documentation
• Academic Paper: "The Design and Implementation of a Real Time Visual Search System on JD E-commerce Platform"