Azure Cosmos DB for MongoDB vCore is Microsoft’s answer to the growing demand for fully managed MongoDB services with dedicated compute power. The service became generally available in 2024 and targets enterprises that want to migrate existing MongoDB workloads to Azure or build new high-performance database applications.
vCore vs. RU-Based Cosmos DB API
Azure Cosmos DB has offered an API for MongoDB for years, based on the proprietary Request Unit (RU) model. This model is excellent for serverless, unpredictable workloads but requires careful capacity planning and differs in some behaviors from standard MongoDB. Cosmos DB for MongoDB vCore resolves these limitations: the service uses dedicated vCore clusters that offer full MongoDB wire-protocol compatibility. Existing MongoDB applications can be migrated to vCore without code changes, as all common drivers, ORMs, and query tools work unchanged.
The sizing model of vCore is immediately understandable for teams already familiar with MongoDB: select vCores, RAM, and storage size by familiar standards. Microsoft handles patching, backups, high availability, and monitoring. Replica sets with automatic failover are available for cluster high availability, enabling the 99.99% SLA.
Native Vector Search and Modern AI Workloads
A strategically significant feature of Cosmos DB for MongoDB vCore is native Vector Search. Many AI applications, especially RAG (Retrieval-Augmented Generation) systems, need both a classic document database and a vector database for embeddings. With vCore, both requirements can be met in a single service: documents and their vector representations are stored directly side by side and queried together in a single query. This reduces architectural complexity, latency, and costs compared to combining MongoDB with a separate vector database.
Scaling and Operations
Compute and storage can be scaled independently in Cosmos DB for MongoDB vCore. Clusters can scale from a few vCores for development environments to large production clusters with many cores and hundreds of GB of RAM. Disk-based storage with automatic expansion avoids manual storage management tasks. Multi-zone deployments are available for high-availability scenarios. The combination of MongoDB compatibility, integrated vector search, and Azure managed service benefits makes vCore the preferred choice for MongoDB migrations to the Azure cloud.
Typical Use Cases
Frequently Asked Questions
What is the difference between vCore and the RU-based Cosmos DB API for MongoDB?
The RU-based API uses Request Units as a capacity model suited for serverless and unpredictable workloads, but requires capacity planning. vCore uses dedicated clusters with fixed vCore and RAM allocation, which is much simpler for existing MongoDB migrations and predictable workloads.
Do I need to modify my MongoDB application?
No. Cosmos DB for MongoDB vCore is fully wire-protocol compatible. Existing MongoDB drivers, frameworks, and tools work without code changes, making migration to Azure significantly easier than with other managed services.
Does vCore support Vector Search?
Yes. Native Vector Search is integrated into Cosmos DB for MongoDB vCore, enabling storage of embeddings directly alongside the actual data. This eliminates the need for a separate vector database for many RAG use cases.
