Just How Persistent Are Your Cloud Bits, Anyway?
Have you ever wondered what happens to your data if you cancel services with your cloud service provider? What happens if you have strict regulatory requirements and policies for archiving and deleting data? The answer is: It depends on the services that you used with the company whether there is any residual data remaining that you need to worry about.
If you did not purchase any backup services from the company and did all backups yourself (through replication, or whatever), then there will be no records or data left. The truth is that the reason cloud companies are reluctant to tell you about their policies on this issue is because they essentially don't back up the data for archive purposes at all. Whatever copies they make (three is the exact number) are strictly for data recovery in case of a hardware failure of the disks themselves.
That is how Amazon Simple Storage Service (S3) works—three for the number of copies of each data object. Therefore, if you delete your data from S3, because there are no snapshots or tapes or any other backups made, the data is gone. If you are extremely paranoid, the bits, of course, do continue to float around on the disks until they are overwritten by a new allocation of some other random company's random data. The VMs are similarly gone once you have deleted them. When you stop a VM in Amazon Web Services, it is permanently gone because it was never on a disk to begin with (VMs run in memory only).
If you purchased backup services, that is a different story. Again, it depends on whether you purchased pooled backup or separate backup. If you backed up your data separately, then you are going to be fairly safe—unless the company is dishonest, which is unlikely. When your contract ended, the data would have been either returned to you or destroyed.
Pooled or shared backups are more problematic. In this situation, your data is backed up along with random other companies’ data. If you want to delete specific records once they had requested to be removed from the system, it is extremely difficult to do because the backed up data is comingled. Pulling out data if it were subpoenaed would be difficult but not impossible to extract from the mass of other companies' backups. Something similar happened with the Megaupload case a few years ago.
Can the disposition of the data be audited? Unfortunately, the cloud certifications that do exist are remarkably toothless. Most of them—if you read between the lines—are just self-audited, and they are focused on data center security and function, not cloud or data lifecycle management. CloudBlue is a company that specializes in creating auditable documentation of data disposal, but they can’t do it for data stored off premises. The bottom line: If you are putting your data in the cloud, there is a small risk that it could be accidently destroyed, but there's a higher risk that it will persist when you want to destroy it.
