Database Development

VST17 Data Modeling and Partitioning Patterns in Azure Cosmos DB


4:15pm - 5:30pm

Level: Intermediate

Leonard Lobel



Sleek Technologies, Inc.

For many newcomers to Cosmos DB, the learning process starts with data modeling and partitioning. How should you structure your model? When should you combine multiple entity types in a single container? Should you denormalize your entities? What's the best partition key for your data?

In this session, we discuss the key strategies for modeling and partitioning data effectively in Cosmos DB. Using a real-world NoSQL example based on the AdventureWorks relational database, we explore key Cosmos DB concepts—request units (RUs), partitioning, and data modeling—and how their understanding guides the path to a data model that yields the best performance and scalability. Attend this session, and acquire the critical skills you'll need to design the optimal database for Cosmos DB.

You will learn:

  • Translate traditional relational database concepts to modern NoSQL data modeling techniques
  • Understand the tradeoffs in embedding vs. referencing, and normalizing vs. denormalizing
  • How to develop the optimal database in terms of scale, performance, and cost