CosmosDB and Avro formats are now available natively in Data Factory's Data Flows. Use Mapping Data Flows with your hierarchies, arrays, and other complex data types to generate and execute data transformation at scale. Full transformation capabilities are supported: aggregations, pivots, joins, calculate columns, etc.
Dynamically scale your ETL process with throughput read and write options.
CosmosDB also includes dynamic throughput scale-up during ETL processes. Avro, JSON, and CosmosDB support hierarchies and complex data types that you can build from the ADF Data Flow browser UI.
Then, you can mash that data up with other lake and database sources in Data Flows in ADF that leverage joins, aggregations, window analytics, and other transformations. Full support for upserts & deletes in CosmosDB is also supported.