The perils of cycling your backupsWe had a disturbing incident recently with a customer whose genealogy database was found to be corrupted. When the customer attempted to view a particular person in the database, the program crashed. Of course our software did not corrupt his database (it is read-only), we simply discovered the fact. The corruption was most likely caused by an obscure bug in the database software, and a rare combination of circumstances.
Our Trellis chart reads every person, event and relationship in a family history database. We stumbled upon corrupted data in this particular case. The original program is no longer supported.
The problem is that this error was probably introduced months or years ago. It was in a rarely-visited corner of a large, 140,000-person database. All of the customer's backups were recent, and they all contained the error. This was a serious problem. Even the software's integrity checking feature terminated abnormally (crashed). There was no way of recovering the data using the program. A GEDCOM export was not possible, as the program crashed when it attempted to export the record in error.
"Cycling" your backups means using the same media in rotation, such as external hard drives, or USB flash memory ("thumb drives"). For example, you use:
On the fourth day of the cycle, you are back to Disk 1.
The advantage of this system is that it is economical. If you accidentally delete a file and catch the mistake at the time it is made, you can go back to yesterday's backup and retrieve a copy.
However, if you do not notice the error (as was the case with our customer), by the time you have cycled through all your backups, the original has disappeared. Your "Cloud" copy is also corrupted. Increasing the cycle to ten or twenty disks (a significant investment) does not solve the problem.
The solution I recommend is to backup to CD or DVD, with a checksum. The CD/DVD will last many years, and will be available to restore the last good copy of your genealogy database, long after you have discovered the error. Detailed instructions are available in an earlier post.
As an alternative, you can exercise the data integrity check of your software, if such a feature is available. The integrity check needs to be done daily.
The story has a happy ending: another program, Ancestral Quest, was able to recover the data.