{"id":7674,"date":"2026-09-29T08:00:22","date_gmt":"2026-09-29T15:00:22","guid":{"rendered":"https:\/\/devblogs.microsoft.com\/azure-sql\/?p=7674"},"modified":"2026-09-28T08:57:01","modified_gmt":"2026-09-28T15:57:01","slug":"diskann-vector-index-search-are-now-generally-available-in-azure-sql","status":"publish","type":"post","link":"https:\/\/devblogs.microsoft.com\/azure-sql\/diskann-vector-index-search-are-now-generally-available-in-azure-sql\/","title":{"rendered":"DiskANN Vector Index and Vector Search Are Now Generally Available in Azure SQL"},"content":{"rendered":"

Today, we are announcing the general availability of DiskANN Vector Index & Search across Azure SQL Database<\/strong>, Azure SQL Managed Instance with the always-up-to-date update policy<\/strong>, and SQL database in Microsoft Fabric<\/strong>.<\/p>\n

DiskANN<\/a> brings scalable approximate nearest-neighbor search directly to the SQL engine. Developers can store vectors alongside relational data and combine vector similarity with the filters, joins, security policies, and transactional data their applications already rely on.<\/p>\n

Why Vector indexes matter<\/h2>\n

Applications use embeddings to represent the semantic meaning of text, images, products, documents, and other content as vectors. Vector similarity search can then find results that are conceptually related to a user\u2019s query, even when they do not contain the same keywords.<\/p>\n

Without a vector index, exact vector search calculates the distance between the query vector and every qualifying vector. It must then identify and order the closest matches.<\/p>\n

\"Embeddings<\/a><\/p>\n

This approach can work well for smaller datasets or workloads that require exact results, but its computational cost increases as the number of vectors grows.<\/p>\n

\"Exact<\/a><\/p>\n

 <\/p>\n

Efficient similarity search with DiskANN<\/h2>\n

A DiskANN vector index provides an approximate nearest-neighbor search path designed to identify relevant candidates efficiently as the dataset grows.<\/p>\n

Instead of evaluating every vector, the index builds a navigable graph so that the database engine can navigate toward the most promising candidates. This reduces the amount of work required to identify similar results while maintaining high search quality.<\/p>\n

\"A<\/a>
A DiskANN vector index helps the database engine navigate efficiently toward the most relevant candidates.<\/figcaption><\/figure><\/p>\n

Why is it called DiskANN?<\/h2>\n

The way I like to remember it –\u00a0 ANN means <\/span>don\u2019t<\/span> search everything, and Disk means <\/span>don\u2019t<\/span> keep everything in memory<\/span>\nMore specifically the name DiskANN<\/strong>\u00a0describes two important aspects of its design:<\/p>\n