Introduction

Azure Cosmos DB for MongoDB vCore is now offering the General Availability of Vector Search. This feature allows users to store, index, and query high-dimensional vector data directly within Azure Cosmos DB for MongoDB vCore, providing vector similarity search capabilities without the need to transfer data to other alternatives.

Create a vector index

Azure Cosmos DB for MongoDB vCore supports two types of vector index algorithms: IVF (Inverted File Indexes) which partitions vectors into clusters and assigns each vector to its nearest cluster center. Users can adjust the latency or accuracy performance by tuning the parameters of the vector index and select a similarity metric (cosine, Euclidean, and inner product).

Perform a vector search

Once data is inserted and a vector index is defined, users can perform a vector similarity search against a targeted query vector. The top k most relevant items in the collection will be returned along with a similarity score indicating the closeness to the query vector.

Integrate with LLM orchestration tools

Azure Cosmos DB for MongoDB vCore integrates with LLL orchestration tools such as Semantic Kernel, LangChain, and LlamaIndex. This integration provides more flexibility in developing applications, enabling lightning-fast query performance and data retrieval.

Integration with Azure OpenAI Service On Your Data

Azure Cosmos DB for MongoDB vCore also offers integration with Azure OpenAI Service On Your Data. This integration allows users to easily use their own data in Azure Cosmos DB for MongoDB vCore with Azure OpenAI completions models. Users can take advantage of the power of OpenAI large language models (LLMs) such as GPT-4 to streamline proof-of-concept development, experimentation, and deployment of web application chat powered by Azure OpenAI completions model.

Resources

To get started with Vector Search in Azure Cosmos DB for MongoDB vCore, users can refer to the official documentation, clone or fork the samples repository, explore the End-to-End RAG Pattern solution for MongoDB vCore with HNSW support, and learn more about vector embeddings with Azure OpenAI Service.

Azure Cosmos DB offers a fully managed NoSQL and relational database for modern app development with speed, availability, scalability, and support for open-source PostgreSQL, MongoDB, and Apache Cassandra. Users can also explore the free tier of Azure Cosmos DB for MongoDB vCore to get started.