Cost-Effective, Low Latency Vector Search with Azure Cosmos DB

TL;DR

This paper presents a cost-effective, low-latency vector search system integrated into Azure Cosmos DB, achieving high performance and scalability by embedding DiskANN, and outperforming specialized vector databases in cost and latency.

Contribution

It demonstrates how to embed a state-of-the-art vector index within a cloud-native operational database, combining high performance, scalability, and cost-efficiency.

Findings

< 20ms query latency over 10 million vectors

43x lower query cost than Pinecone

Scales to billions of vectors via automatic partitioning

Abstract

Vector indexing enables semantic search over diverse corpora and has become an important interface to databases for both users and AI agents. Efficient vector search requires deep optimizations in database systems. This has motivated a new class of specialized vector databases that optimize for vector search quality and cost. Instead, we argue that a scalable, high-performance, and cost-efficient vector search system can be built inside a cloud-native operational database like Azure Cosmos DB while leveraging the benefits of a distributed database such as high availability, durability, and scale. We do this by deeply integrating DiskANN, a state-of-the-art vector indexing library, inside Azure Cosmos DB NoSQL. This system uses a single vector index per partition stored in existing index trees, and kept in sync with underlying data. It supports < 20ms query latency over an index spanning 10 million vectors, has stable recall over updates, and offers approximately 43x and 12x lower query cost compared to Pinecone and Zilliz serverless enterprise products. It also scales out to billions of vectors via automatic partitioning. This convergent design presents a point in favor of integrating vector indices into operational databases in the context of recent debates on specialized vector databases, and offers a template for vector indexing in other databases.