

Most of these backup solutions try to reduce the amount of data backed up, so the system will process each file and determine if the contents changed and only upload new data. In this case if you modify the file you may end up with 2 16 MB chunks that have slightly different data, and only when the old backup expires and the pruning process completes would one of those chunks be deleted.
Smaller chunks sizes means you get more files written to S3, but less duplicate data when deduping without compression.

On the flip side of this, any reasonably sized environment should be running their own ntp pool for their servers to use. Being down one stratum is usually not that big of a deal for most environments.