Before he went home to redesign the world’s most famous home office, the Ur-Guru and I visited the Computer History Museum in Mountain View. In addition to watching a demonstration of Babbage’s Difference Engine #2 and sitting down to rest on a Cray supercomputer (and of course pointing out which of the early personal computers each of us had owned), we saw one of the first ever hard drives. It was bigger in diameter than the tires on my car. There’s actually a project to restore the IBM RAMAC, the computer that used it, over at the Magnetic Disk Heritage Center. (Before disk drives, computers used something called core memory. I was pleased to discover that the Ur-Guru didn’t know what that was, either.)
But while all that history is fascinating, and I recommend a trip to the museum if you get the chance to go, it won’t help you preserve your data today. So I’m introducing today’s guest blogger, Network World columnist James Gaskin, who offered me this article about “Pirate Backup” (originally published in 2007, but still relevant).
In honor of the upcoming Pirates of the Caribbean movie, let me introduce the Pirate Backup System (ARR). ARR, besides a bit of pirate talk, stands for Automatic, Redundant, and Restorable. Meeting those three goals makes a good backup system. Meeting only two will lead to disappointment. Meeting only one describes about two thirds of small businesses today.
I’m working on a “How To Fix IT Manual” called Data Safety Using the Pirate Backup System (ARR). But focusing on backup sends the wrong message, because backing up files does nothing. Users only get value when they restore files. Backup is just the necessary pain to reach the gain of restoring important data files when they are lost, stolen, or mangled. Think of backup as the insurance premiums, and restoration as the replacement payment after a loss.
The first A in ARR, Automatic, forces businesses to take into account the point made in the previous paragraph: users get no value from backup. Hence, users don’t back up. One may consider this short-sighted, but users will complain that participating in any backup procedures means they’re doing the administrator’s job, not their own. Since data safety is our goal here, let’s not argue with the users about this today, let’s just work around those issues.
To be automatic your backup system must work without any user intervention. You can’t even trust users to leave their computer turned on for a backup job to run at 3:00 a.m. You certainly can’t trust them to click an icon to run a backup.
These restrictions leave two options: data must be saved somewhere besides PCs, or you must place software on each computer that works without user intervention.
The first option, the neatest, stores all user data files on some type of centralized file server or even a shared online workspace. That would be great, but realistically you will need to install software on each computer to back up data files on a schedule or immediately upon every file change. Almost every backup software application will schedule backups on at least an hourly basis. To grab file changes immediately you will need special software from the Continuous Data Protection (CDP) range of products.
Second, redundancy protects data files, and makes disaster recovery possible. Companies learn the hard way that backup tapes sitting beside the server burn up when the server burns up. They also learn backup network-attached storage devices get stolen when thieves steal their servers. You must keep copies of data files somewhere outside your business to recover from a wide variety of disasters small and large.
Back when tape cartridges led the backup media world, people developed offsite tape rotation schedules using a Tower of Hanoi algorithm to try and keep the right data on the right tapes at the right places. Those never worked, because people dropped the ball quickly.
Today you can send data offsite much more easily than before. Many service companies offer excellent prices to accept data files across the Internet at their remote data storage center. You can send data from one office to another office, or send data files to a hidden directory on your company Web server. You have multiple options, but you need to pick one or two and get started. If you try to carry USB hard drives back and forth from your servers, however, you will soon get tired and quit.
Finally, files must restore properly or you’ve wasted all your time. Good backup software makes it easy to restore files to their original location or other locations. Offsite services with file redirection make it easy to share files between remote locations, but be careful that file versions don’t get changed by one user without the other users knowing. But a good restoration test is to pick a data file folder at random, restore it to another location, and check those files.
If you want “bare metal restore” capabilities to quickly rebuild a personal computer or server, you’ll need special boot CD-ROM disks tied to the backup files. Each backup vendor offers different methods of bare metal restore, but you can keep those boot disks close at hand even if the data stays safe in an offsite location across the country. But this example points out the need for several options in your backup system, and why I call it a system with multiple parts rather than a backup method or backup process.
Remember, you can’t trust users to help you perform any backup chores. You can trust the Pirate Backup System, however, especially if you put a parrot on your shoulder before you say ARR.
Copyright © 1994-2008 Network World, Inc. All rights reserved.
Although I’m now minus one major source of distraction, I have a lot of client work to catch up on, and I got so many responses to my call for volunteers to write guest posts that I could go on for at least another month without contributing anything original. I promise to write at least one post per month for myself, however!