The Package service stores information about all of the packages. The storage requirements for the Package are:

Long-term storage.

Able to handle a high volume of packages, requiring high write throughput.

Support simple queries by package ID. No complex joins or requirements for referential integrity.

Because the package data is not relational, a document-oriented database is appropriate, and Azure Cosmos DB can achieve high throughput by using sharded collections. The team that works on the Package service is familiar with the MEAN stack (MongoDB, Express.js, AngularJS, and Node.js), so they select the MongoDB API for Azure Cosmos DB. That lets them leverage their existing experience with MongoDB, while getting the benefits of Azure Cosmos DB, which is a managed Azure service.

https://www.turing.com/blog/microservices-data-management-patterns/