Vald

Vald is an open-source, highly scalable distributed vector search engine designed for fast approximate nearest neighbor (ANN) search on dense vector data. Built on a cloud-native architecture, Vald enables efficient, real-time large-scale vector search operations.

Features

• Highly Scalable Distributed Architecture: Designed for horizontal scaling across memory and CPU, capable of handling billions of feature vectors.
• Fast ANN Search: Utilizes the NGT (Neighborhood Graph and Tree) algorithm for efficient approximate nearest neighbor searches.
• Asynchronous Auto-Indexing: Supports non-blocking, distributed auto-indexing, allowing the system to remain operational during indexing processes.
• Automatic Index Backup: Offers auto-backup capabilities using object storage or persistent volumes for disaster recovery.
• Distributed Indexing: Distributes vector indices across multiple agents, each storing different parts of the index.
• Index Replication and Rebalancing: Stores index replicas on multiple agents and can automatically rebalance replicas if an agent fails.
• Customizable Ingress/Egress Filtering: Provides configurable filters for data input/output, adaptable to gRPC interfaces.
• Multi-language SDKs: Supports SDKs for Golang, Java, Node.js, and Python.
• Highly Customizable: Allows configuration of vector dimensions, number of replicas, and other advanced settings.
• Easy Installation: Designed for quick and simple deployment.

Category

• Vector Database Engines

Tags

• open-source
• distributed
• scalable
• real-time

Pricing

No pricing information was provided; Vald is open-source.