Disaster Recovery

Four tips for preserving digital data

I've written at length about the difficulties associated with preserving digital data for the long term. To round it all off today, I've condensed a number of tips to keep in mind should you find yourself with the unenviable responsibility for storing data.

1.  RAID is not fail-safe

I've said it before, and I'll say it again: RAID is not 100% reliable. Yes, by itself, RAID does dramatically increase the survivability of your data. However, anecdotal evidence points to the fact that simultaneous disk failure is not impossible or even rare.

To enhance your chances with RAID, ensure that you set up your storage cluster with the right mix of hard disk drives selected from different manufacturing batches. Configuring each RAID array with at least one hot-failover drive -- available on higher-end RAID controllers -- is a minimum in my book.

In addition, it's imperative to assign someone who will check it periodically -- preferably as part of their daily routine. It might sound superfluous, but it's very possible for a hard drive in a RAID configuration to fail for a while before anyone notices.

Some RAID drivers come with a front-end that allows you to define alerts or e-mails to be sent in the case of failure. Your mileage might vary though, and it's always good to physically check whenever possible.

And of course, you should have one additional copy of the data at the minimum, be it near-line, off-line, or off-site.

2.  You will have to factor in a media reader or interface

One aspect that gets overlooked in regard to storage is the physical reader required to read the storage medium, or, in cases of hard disk drives, the availability of the hardware interfaces to connect to them.

It's not hard to ensure that tape media are properly stored in a humidity controlled environment, or that your SCSI drives are properly stowed away under lock and key. But can you be certain that you'll be able to read the data safely archived in your backup media of choice should the need arise? Is your tape drive or Iomega Zip drive still working?

Retaining the long-term ability to read older media is one of the chief reasons tape drive companies develop multi-year roadmaps for their tape technologies.

3.  Archiving data costs money

You should realize from the start that it costs money to store data. Be it near-line hard disk drives that require electricity to power, or inert tape drive cartridges that require humidity controls, there will be a cost. So you must decide what and how much data needs to be archived, and allocate an appropriate budget for it. And management must understand this from the start.

4.  You will have to have a plan for defending against internal sabotage

Admittedly, this probably has more to do with the rigorous implementation of data protection practices and policies. Depending on the corporate culture and the decision makers in your company, this might introduce policies that can be easily implemented -- or are outright impossible.

The fact is that it's next to impossible to thwart a determined saboteur who really wants to cause the maximum possible damage. But that's no excuse for implementing the barest policies or safeguards.


Paul Mah is a writer and blogger who lives in Singapore, where he has worked for a number of years in various capacities within the IT industry. Paul enjoys tinkering with tech gadgets, smartphones, and networking devices.


I have been in two different seats on this. First, I was an operator on a system that crashed. We rebuilt it, and asked to get a restore of our data. Oops, the backups aren't restoring from tape. Looks like our tapes are worn out but we never notices. Second, I was a manager of a group of DBAs. One of the things I did every quarter, as ask for a random database, from a tape that had gone "off to the mountain" to be returned and restored. People complained about this. It was extra work and they were busy. I even had to have manager to manager discussions about this, that finally came down to. "My group needs to have a db backup, that is on tape restored. This is an official request, the we are placing on your support ticket. Our company audits the response time to these tickets and your teams performance will affect your teams bonus." We started getting our backups restored right away after that.


My first experience with losing data was back in the late 80's with my contact list. It was my only copy of all my friends, and I regularly backed it up to two tapes. Unfortunately, I never did a restore and one day I lost the data. Not to fear, I had 2 backups. Restore 1 failed, but I still had the other. Very bad feeling when the second one didn't restore. Since then, testing your backups once a month is better than a very bad feeling. You don't have a backup if you can't restore it.


Not just get it back, but get back what you want. As part of our year-end IT housekeeping, we just went through a set of (verified) backup tapes from our factory floor network, and received a nasty little surprise. The set would not restore an individual file or folder. Now there's a message I never want to see again - "The tape device reported an error on request to read data: Hardware Failure" Ironically, the error had nothing to do with hardware, as I could easily restore the ENTIRE set... Needless to say, I have modified our backup strategy. We now use data images on hot swap scsi drives, and have access to any portion of any backup at any time. Lesson learned: Don't just verify the data, verify your ability to access the data!


Backups are the basic first step. A no-brainer even. Restore is the effective test of any backup policy. I use tape as the main medium (200/400gb) but also carry out daily DVD backups for financial data, weekly DVD backups for user data, and two-weekly backups for changed policy data. Yes it's extra work, but when the main tape unit mangled 3 days backup tapes, the DVD's came to the rescue. And I was flavour of the week, as my DVD backups weren't company policy at the time. Since then, they have been. However, any backup plan must be suitable for the organisation. The above won't work for large organisations with 1000s of users.

Editor's Picks