Single page Print

Toward a more rational backup strategy

It's easier—and cheaper—than you might think
— 10:43 PM on July 7, 2009

In the week leading up to the birth of our new baby boy, I was in something of a holding pattern, unable to start on a major new project, yet compelled to continue working on something to pass the time. I took that opportunity to do something I've put off for years in an epic feat of procrastination: the formulation of a sensible and fairly comprehensive backup strategy. Like many of us, my history here is a shameful one, a checkered past littered with poorly labeled DVD-Rs, manual file copies, and a minimum of forethought. Yes, I have been backing up my most critical files for years, but not in a way that makes a heck of a lot of sense.

This is a bit of a painful admission for me, in part because I come from a past life as a professional network and system administrator, in which I successfully devised and managed complex backup schemes across entire data centers. But mostly the admission is painful because backups are one of a few subjects in computers—along with network security—where a certain class of geek enjoys lecturing people stridently about Best Practices and acting as if they have Special Enlightenment Knowledge about how these things should be done, as if ensuring you don't lose your personal photos were akin to a moon shot. Admitting anything less than perfection on this front could lead to a loss of geek accreditation, among other things.

This is a risk I am willing to take for you, gentle reader.

Fortunately, I was surprised to see how little effort it took to arrive at a much more rational backup scheme, one that encompasses virtually all of my critical data and is, by and large, automated in software. It was cheap to institute, given the hardware and software I already had on hand. And honestly, the scheme I've set up feels outright competent and reassuring. I'll share my experiences below, so that they might help you, if you've not yet devised a means of managing and backing up your data that seems satisfactory.

The right mix of hardware
I should say that I started with pretty much all the necessary hardware on hand, since my intent to do the right thing preceded the actuality by a number of months. Late last year, I purchased a bundle of storage hardware aimed at addressing this problem: a pair of 1TB Caviar Green hard drives, a Thermaltake BlacX external USB drive dock, and a Barracuda 7200.11 1.5TB drive. Today, all of these items together can be had for under $400, and they constitute pretty much everything you need for a really solid storage and backup solution.

You may notice here the lack of an optical drive, such as a Blu-ray burner. Time and experience have convinced me that removable optical storage isn't a particularly convenient or cost-effective way to handle backups, at least for me. I'm madly demotivated by disk swapping, and hard drive capacities have outpaced optical media by leaps and bounds. I think most folks will be better off using one or more hard drives for last-line data storage, most of the time. I do have DVD drive and recommend using it as a part of a backup scheme, but only in a small role.

In any event, equipped with this hardware, I set up a storage hierarchy that looks something like this: at the top is the 80GB Intel X25-M SSD that holds my OS, applications, email, and the like. The second tier is a terabyte RAID 1 array comprised of those two Caviar Green drives. I had originally planned to host this array in my local Windows XP-based storage server, but some strange file sharing incompatibilities between XP and Vista torpedoed that idea. Instead, I popped those drives into my main PC, the Damagebox Quad, a move that turned out to be fortuitous, as you'll see. This array is then shared over my local network with other machines. Tier three is that 1.5TB Barracuda, which I plug into the BlacX dock in order to make backups. The rest of the time, the 'cuda stays unplugged and safely hidden, impervious to power surges, lightning strikes, or other such drive-frying calamities.

A surprising software option
Having this basic hierarchy in place was nice but far from sufficient. I was still using a combination of individual backup programs, such as one built specifically for Windows Mail, and good ol' file copies to manage my data. If I didn't sit down every so often and go through the motions of copying files to the offline drive, I wasn't much better off than before. Yes, most of my data were on a RAID 1 mirror, so I was insulated from a single drive failure, but as you may know, a RAID array is no substitute for a proper backup. And I had other critical data on my wife's PC and the storage server that weren't even on this RAID array. To name just a few of many problems.

After looking into what sort of backup software I might want to purchase to help straighten things out, I decided to first try using Windows Vista's own built-in Backup Center feature to see whether it could meet my needs. Backup Center has the great virtue of being already paid for by my Vista license fee, and heck, I figured I might get a little bit of return on my investment in Vista Ultimate, which comes with some additional backup capabilities.

To my great surprise, Vista's backup features turned out to be more than acceptable for my needs, even though its two basic backup functions are very simple to use.

One of those, dubbed Windows Complete PC Backup, is only available in Windows Vista Business, Ultimate, and Enterprise editions. This feature works much like Norton Ghost, creating an image of your entire system, including data, programs, and preferences. In the event of a catastrophic failure of primary storage, one can boot from a Vista DVD and use the included tools to restore an entire system to its prior state on a new drive in a matter of minutes, with essentially everything intact. One may specify which drives to include in the complete backup. I included only my 80GB primary SSD, since my system boots from it and can function without any other drives attached. The compressed image occupied only 54GB on a mechanical drive, a small price to pay for such security. Complete PC Backup runs in the background from inside Windows, no reboot required, and once you have a full image, the utility can save incremental backups from there, storing only the changes since the last complete backup.

A backup of all of your data is really no substitute for this sort of disk image, which brings real peace of mind that you won't have to spend a few days restoring your carefully calibrated PC to its former state, should the worst happen. If your OS doesn't include Complete PC Backup, one may instead use a third-party program like Ghost or Acronis TrueImage to achieve the same result.

As far as I can tell, though, Complete PC Backup cannot be scheduled to run automatically via a GUI (though it is possible via the command line), and one can only restore the complete image to disk; individual files can't be restored piecemeal. Thus, it's not an appropriate stand-in for a separate backup of one's user data. For that job, Vista has a separate file backup function.

The setup wizard for Vista's file backup is off-puttingly simple. The user may choose which drives to include in the backup, and then you're prompted about what file types to include. At first blush, I was pretty well convinced this utility was overmatched against my thorny, years-in-the-making collection of various file types in different locations across multiple drives. On reflection, though, this consumer-friendly, data-oriented approach makes perfect sense.

You can see the choices I made in the dialog above. Since I still use Windows Mail, the check-box for "E-mail" should have me covered there. "Documents" is more expansive than it may sound, because it covers all file types registered as documents with any installed programs, in any location, not just the "My Documents" folder or what have you. "Compressed files" covers things like Zip archives, which I definitely want to keep, and "Other files" includes unrecognized file types—so just about anything that isn't an executable or system file. Checking this last box perhaps wasn't necessary to cover all of my critical data, but it made me feel more comfortable with how the program works. The only file type I've elected to exclude is "TV shows," presumably ones recorded via Windows Media Center, since they take up lots of space and can generally be replaced via, ahem, other channels.

The wizard then prompts you to schedule this backup job. I chose to run mine weekly, on Friday afternoons at the end of the work week. There is still an element of manual intervention, since I'll have to pop the 1.5TB 'cuda into its cradle before the backup runs, but otherwise, everything just happens automatically, running in the background with nothing but a little notification icon in the system tray.

Interestingly enough, Vista's file backup stores its data in a series of large, compressed Zip archives. Although it's a cumbersome way to find a file, one may browse through them after changing their permissions to allow it. The file backup does store incremental backups, but a full copy of any file modified will be included in each backup, so it's not particularly efficient. Since my storage capacity exceeds the size of my total data set several times over, I haven't found this limitation to be troubling.

In fact, inspired by the logical clarity of Windows' data-centric backup approach, I finally gave in a little bit and mapped my custom directory structure for my main data, formed over years of personal and professional use, to Windows' main categories for Documents, Pictures, Music, and the like. By no means would all of my files fit in Redmond's suggested, default locations on the C: drive, which is my 80GB SSD. Instead, I changed the properties for my profile's main data directories, pointing them to the locations where I keep my files of the appropriate type. In the image above, I've mapped "Downloads" to a subdirectory on my terabyte RAID 1 array. Now, the file selection shortcuts in the OS actually work, because they're mapped to the appropriate places. Making these changes also allowed me to move my "Documents" folder to the RAID mirror, freeing up crucial space on the SSD and placing those important files where they really belong.

Scheduling a weekly backup job had my PC more or less covered, but what about the other PCs on my network? Well, I used the built-in Windows file backup on both of those systems (Vista on one, XP on the other) to schedule weekly backups over the network to my RAID 1 mirror. That way, if either of those systems goes belly up, I have their data preserved. Neither of those systems has a particularly large set of critical data, so I'm not hitting any space constraints. And those backup jobs are set to run earlier in the week, so when my PC is backed up to the external 1.5TB on Friday, a recent copy of their files is included.