Deploy 10,000 IoT temperature sensors globally and require real-time ingest, processing, storage, and analysis at scale. What architecture should you use?

This architecture leverages Google Cloud's fully managed, scalable services for real-time data processing. Pub/Sub provides a highly scalable messaging service for ingesting the 10,000 global sensor data streams. Dataflow, a serverless stream processing service, can handle the real-time processing and transformations required, scaling automatically to meet demand. BigQuery, a serverless data warehouse, is ideal for storing and analyzing the processed data at scale. Ingesting into Datastore and then exporting to BigQuery introduces unnecessary latency and complexity for real-time needs. Writing raw telemetry to Cloud Storage and using on-demand Dataproc is a batch-oriented approach, not suitable for real-time requirements. Exporting logs in batch to Cloud Storage and importing into Cloud SQL is also a batch process and Cloud SQL is not designed for the scale and analytical queries of this type of IoT data.