diskann

DiskANN is a disk-based approximate nearest neighbor search algorithm that enables high-performance vector similarity search directly within SQL Server and Azure PostgreSQL. Recent discussions highlight its integration into SQL Server 2025 RC0 for native vector search capabilities, allowing on-premises databases to support AI workloads without external services. In Azure PostgreSQL, DiskANN powers in-database AI features, enabling efficient vector indexing for semantic search and recommendation systems. The algorithm is designed for large-scale datasets, balancing speed and accuracy while running on standard hardware. These developments position DiskANN as a key technology for bringing vector search closer to relational data, reducing latency and operational complexity for AI applications.