Comment by PaulHoule

This happens in AWS too. In the cloud you have to assume nodes will go down and not have it be a disaster if one fails.