The most obvious liability is in wading through old data looking for items to satisfy an e-discovery request. When an employee brings a harassment suit claiming that your company has created a hostile work environment for 20 years, the first thing any attorney will do is subpoena 20 years of emails and any other digital data, hunting for evidence.
If you have 20 years' worth of old backup tapes, the court may--nd likely will--require you to scour them for relevant data. And judges don't want to hear excuses; courts have generally been unsympathetic to arguments that recovery would be onerous because backup software no longer maintains a catalog of the data on the tapes, or that IT no longer has the DLT or DAT drives needed to read them. Cue the expensive consultant.
If, however, you have in place a five-year retention policy for email and encryption of data at rest, you're in much better shape. Set your backup application to encrypt email backups and change the encryption key once per year. Every year, delete the encryption key for all data that has now turned six, and that information--even if it shares tapes with other data that has different retention requirements--is effectively destroyed. No key, no smoking gun email.
Another good application for this technology is simplifying disposal of enterprise storage systems and their constituent disk drives. When a storage system reaches the end of its useful life or, worse, dies, IT has to figure out how to wipe the data before the system it can be disposed of or resold.
While hard disks can be overwritten with fresh data, this takes time and, of course, requires that the drive actually be working. Many data centers have piles of old or failed hard drives sitting around, waiting for a good disposal process, like disk shredding or degaussing.
The whole process is further complicated with systems and devices, like SSDs, that dynamically map logical to physical blocks. A typical 200-Gbyte enterprise SSD actually has 256 Gbytes or more of flash memory plus a flash controller chip that dynamically maps logical disk blocks to flash memory pages as data is written. Without detailed knowledge of how the particular flash controller in a given SSD performs that mapping, there's no way to be sure that every flash page in the device has been overwritten--regardless of how much data you write to the drive.
If your organization standardized on self-encrypting drives, however, used disks could not only be safely disposed of, they could be readily reused because any data they contain would be unavailable without the encryption key. Array controllers would keep the encryption keys for each drive in flash or other nonvolatile memory and pass them to the encryption chip on each drive at power up. When a drive is removed from the array, its data is unavailable because the drive has no access to the encryption key. That's so whether the drive works or not, and regardless of whether the folks who want your data are willing to use an electron microscope to read it.
IT can use this digital shredding technique now, with some effort. The risk, of course, is that you'll lose a key and be unable to access encrypted data that you do need; key management is still a trial. But security pros like encryption, so it shouldn't be difficult to get help setting up a management system.
The next step is for the makers of object storage systems to build digital shredding into their policy engines. A compliance archive using digital shredding could assign an encryption key to all data objects whose retention period expired on a given day and then automatically discard the key when the expiration day arrives. Keep an eye out for this feature.