Ask HN: How does S3 locate object data?

What's the average amount of data in a region now? Exabyte? Jumanjibyte? How do they locate a specific object in a literal sea of harddrives?