A web service must ingest and store 500,000 requests per second, support real-time exact-match queries on known attributes, have idle periods, and be cost-sensitive. Which compute platform and database pair should you use?

Cloud Run is ideal for this scenario because it scales to zero during idle periods, significantly reducing costs, and can handle high request rates by rapidly scaling up. Cloud Bigtable is a wide-column NoSQL database designed for high-throughput ingestion and low-latency lookups on massive datasets, making it perfect for 500,000 requests per second and real-time exact-match queries. BigQuery is an analytical data warehouse, not optimized for real-time exact-match queries on individual records, making it unsuitable. Compute Engine autoscaling managed instance groups offer flexibility but incur costs even when scaled down to a minimum instance count, making them less cost-effective during idle periods compared to Cloud Run's scale-to-zero capability.