Building a distributed Key-Value store to power S3

An interesting snippet from the linked paper:

  At the time of writing, Hive has been in production on the Scaleway S3 product for around a month. In that time, 5 billion entries have been written, with an average write P90 around 1ms and an average read of P90 around 150us. The aggregate availability (return_code != 5xx / total_requests * 100) is 99.9998%,