Would you happen to know the model that was used to make those estimates? It seems to me that I could live with a 100 year mean time to data loss. How do the mean time to data loss change with the number of drives?
Or does anyone know how using AWS EBS effects this calculation?
I'd ascribe the meantime calculation based more on the probability of SPs getting affected when you have more disks under a fewer SP umbrella when the same is compared to fewer disks under a large number of SPs.
To my knowledge, this calculation should be platform agnostic and hence should not make any difference whether the setup is on-premise or cloud-based. Of course, you could have interfering parameter(s) like the reliability and durability of the so-called platform and components used that may affect the theoretical calculations.
Retrieving data ...