I always just see if I can restore a file from the backup. If it isn’t there in the first place or you can’t retrieve it, you have a problem! Checks should probably be performed on a schedule since backup mechanisms can stop working.
I use the ‘Backup and Validate’ feature in Arq backup to validate against my external drives. Arq also runs a validation automatically every 30 days (this may be an option to enable/disable). Not sure if it works with cloud backups, but it sounds like it does:
1 - Every X months, weeks, or days (depending on state of paranoia) verify backups:
a - Verify that everything looks like it’s there.
b - Restore some files of different types, especially including some of the more important files
2 - Every X months (less frequent than above):
a - Verify using one of the methods from the Arq article (above)
b - Verify using one of the methods from the CCC article (above)
I’ll use Apple Reminders to prompt me to do the verifications.
Thanks to all for your suggestions and to @JohnAtl for the links to the articles.
The largest amount (in terms of storage) is my photo collection: if one or two of the tens of thousands of photos is lost, from ALL backups, the damage is still negligible. Honestly, I don’t really check it. The backup itself checks checksums, good enough for me.
My most vital data: financial records, contracts, invoices, documents, published work, scanned stuff … is slim by comparison. So: easy enough to restore the whole folder/sub-folders once in a while.
So far, much of the discussion has been about restoring files and giving a visual check that they’re there.
A more thorough test would be to restore a set of files to, e.g. a folder on the desktop, then use a utility like Beyond Compare to compare the contents of the restored files with existing files on the drive. This would help ensure that not only are the files there, but their contents are correct too.
A little more work, and perhaps overkill for some things. It all depends on how valuable your data are.
The ultimate would be a full restore to a spare drive and comparison with the backed up drive.
In addition to the techniques mentioned I like to use disaster scenarios. Come up with events that would cause data loss and see if you can recover from them. An example,is a Paradise CA type fire. If that happened to you, could you get your data back. Another example we used in corporate DR recovery testing was to bring up critical systems like payroll but do it with certain people being unavailable, i.e. dead.