Question

Your company runs an operational Azure Cosmos DB for NoSQL account that backs a retail order-processing app. The BI team wants to run Spark and T-SQL queries over the operational data in Microsoft Fabric with near real-time freshness, without provisioning analytical throughput, writing ETL pipelines, or affecting the transactional workload's RU consumption. They also want the replicated data stored in OneLake in an open Delta/Parquet format so it can be reused by other Fabric engines. Which approach best meets these requirements?

Accepted Answer

Configure Fabric mirroring for the Cosmos DB account, which continuously replicates the data into OneLake as Delta tables at no RU cost to the transactional workload.

Answer

Enable Azure Synapse Link analytical store on the container and connect a Synapse Spark pool, then copy results into Fabric with a scheduled pipeline.

Answer

Build an Azure Function triggered by the change feed that writes each change into a Fabric Lakehouse table via the REST API.

Answer

Use Azure Data Factory to run hourly Copy activities from Cosmos DB into a Fabric Lakehouse in Parquet format.