Vald stands out as a highly scalable distributed vector search engine, specifically designed for fast and approximate nearest neighbor searches in dense vector spaces. Built upon a Cloud-Native architecture, Vald incorporates the NGT algorithm for efficient neighbor searches and offers features that make it both powerful and user-friendly.
Why Choose Vald?
- Asynchronous Auto Indexing: Vald employs a distributed index graph, avoiding the stop-the-world scenario during indexing that is common in traditional graph structures.
- Customizable Ingress/Egress Filtering: The engine features a highly customizable Ingress/Egress filter, adaptable to fit the gRPC interface, providing flexibility in handling data flow.
- Cloud-Native Design: Vald is horizontally scalable on memory and CPU, ensuring it can meet the demands of varying workloads.
- Auto Indexing Backup: Vald supports automatic backup features using Object Storage or Persistent Volume, enhancing data resilience and enabling effective disaster recovery strategies.