A telemetry system must ingest 100 MB/s into Azure Event Hubs from 100,000 devices. Three downstream teams need independent reads: a real-time analytics job, a monitoring service, and a cold-path archive. Data must be automatically persisted to Azure Data Lake Storage Gen2 every 5 minutes, and downstream reads must not interfere with each other. Which two configurations should you recommend?

Creating a separate consumer group for each downstream pipeline (real-time analytics, monitoring, archive) ensures that each team can read events independently from their own position in the event stream without affecting the others. This is a core feature of consumer groups in Event Hubs. Enabling Event Hubs Capture to Azure Data Lake Storage Gen2 with a 5-minute window directly addresses the requirement for automatic persistence every 5 minutes. A single partition would limit throughput and not scale for 100,000 devices. Using the Schema Registry is good practice but not directly required by the prompt's core requirements. Setting the partition count to 2 is insufficient for the specified ingestion rate and number of devices; a higher partition count would be needed for better parallelism and throughput.