Without wishing to tempt providence, over the years I’ve been fairly lucky with my data. The worst time was a few years ago when I lost the whole of my home accounts database due to a hard disk crash (probably self-inflicted – I was trying to find out first-hand about the intricacies of over-clocking).
The ultimately futile experience of rummaging around to find backups of the data changed my life – I stopped doing my accounts on my PC and reverted to what I’d been doing five years previously – throwing my bank statements straight into the bin.
I now make sure that any work data that’s worth keeping is copied somewhere easily accessible, such as a folder on a company file server. After all, this means it’s backed up daily and the tapes are stored off-site in a safe location.
Until recently I’ve never had to test this system. But the other week, a useful, but not vital, shared spreadsheet decided it had had enough and became corrupted. Even Excel’s repair utility couldn’t make sense of it. No problem, I thought, I’ve been a clever chap, using Microsoft’s SyncToy to copy the file over to my local PC every day in anticipation of just this eventuality. But it turned out that the corruption had happened a few days previously and so all I had was a useless backup file on my hard disk. Excel had been set to back up the file as well with every file close, but that backup was also dead.
So I duly asked the nice helpdesk staff to please get me my file back, which they did, and of course it was corrupted as well. As was the previous day’s version. So there ensued a tedious process of recalling tapes until we got a file in an operational state.
That’s the problem with dumb backups, of course. They simply copy the bytes over, with no way of knowing whether those bytes will make any sense to an application. Post-backup verification just checks whether the bytes are readable.
I know there are tools out there, such as Norton’s GoBack, that keep track of all changes to files on a disk and let you revert to a previous version. Lenovo’s Rescue and Recovery works in a similar way. But these are sledgehammers that can pound system resources. They are also limited in the amount of history they can maintain and they aren’t designed for use on networks.
I’m not sure that a backup program could ever be expected to verify file structures. Another approach to this problem would be a plug-in to office applications such as Word and Excel that automatically and intelligently copies a file when it’s successfully closed (or preferably when it’s successfully opened) to a sequentially-numbered backup file or directory of your choice. And ultimately, full versioning and backup control should be built into the application itself. Surely it’s not a lot to ask – if anyone knows of such a tool, I’d be more than happy to hear from you.





reader comments