How would you design a high-throughput data ingestion service with load balancing and database storage choices?
