When weighing whether to go to the cloud or which cloud service to use when you do, it's a no-brainer to make sure the service's reliability, cost effectiveness, and security track records match your requirements. However, many cloud buyers don't carefully consider two other critically important characteristics of any cloud service: the degree to which you can control your services without intervention from your vendor and the degree to which you cede responsibility for day-to-day management tasks to your vendor.
When I think back to some of the more well-publicized failures of megascale cloud infrastructures (Amazon Web Services and Microsoft Azure), each has been followed by enraged forum posts from users who had incurred downtime, lost production data, or both. It's entirely understandable for customers to be irate when a service they're paying for fails. However, it's also an excellent indication that these cloud users didn't fully understand that they still bore a large responsibility for managing and protecting their systems.
Cloud vendors should shoulder some of the responsibility for this disconnect. Although it might not be good for business, big, red warning text suggesting that users are responsible for backing up their data or maintaining replicated failover instances in different availability zones might have dampened some of those anguished cries.