[ddorian43]

I'm too lazy to find the slide deck from AWS CTO where he explained it, the bigger the total scale the easier it was to support bursting workloads by spreading chunks of data across the whole fleet of disks.