“Today, we have tens of thousands of customers with S3 buckets that are spread across millions of drives.”
S3 runs on millions of hard drives, and one genomics customer’s burst from thousands of Lambda functions can land on over a million of them. Amazon’s trick is aggregation. Millions of bursty, idle-most-of-the-time workloads smooth into predictable demand. Your bucket is cheap because it hides in a crowd of everyone else’s.