Restoring Data Isn’t Rocket Science

A blinding light crossed the Texas sky shortly after dawn on 1st February, 2003. Those who knew what it was gasped in horror as the Space Shuttle Columbia burned during re-entry into the Earth’s atmosphere, leaving a trail of debris that stretched from central Texas into Louisiana, and killing the seven crew members on board.

In the debris was a Seagate hard drive containing irreplaceable scientific data. The drive fell to earth and came to rest on a dry lake bed where it would remain undisturbed for six months.

Data from space debris

Eventually someone looked closely at what appeared to be a burned chunk of metal and discovered that it was a hard disk drive. Somewhat later, NASA sent the drive to Kroll Ontrack in Minneapolis, to see if the data could be saved. In a few days the engineers at Kroll did just that, recovering 99 percent of the data.

The event that sent me to Kroll Ontrack was far less tragic, but still important to my work. My adventure with data recovery began in early July when I suddenly was unable to access the Buffalo Linkstation Quad NAS server where I stored copies of photos. I walked into the lab and saw that the power and drive lights were lit, I checked the network cable, and it was attached with the link lights showing a connection. But nothing I could do would let me get to the data on that server.

Of course I did the obvious things, including cycling the power, checking to make sure the drives were firmly seated, but nothing worked. So I called the support number at Buffalo Technologies for help. Unfortunately the company’s product support specialists didn’t provide much help. One support tech didn’t understand what a RAID 5 array was. Another one thought there might be a network problem.

Finally, I got a tech support staffer who at least understood the question, but the news wasn’t good. It seems that this Buffalo NAS device uses an embedded Linux that operates the server and that also handles the drive controller and the LAN interface. The firmware was obviously corrupted, I was told. There is no fix. My only choice was to try to find and buy a server just like the one I had.

Sadly, servers just like the one I had aren’t available in the United States, although I could have imported one from Australia. I was told that it was possible that I could buy a new, similar server and it might work, but that there were no guarantees. The thought of spending $600 on a solution that might work (or might not) was unappealing. So I looked elsewhere.

Finding a data recovery option

First I went to a data recovery company recommended by Buffalo and found that I’d have to ship the server to the West Coast, potentially spend several thousand dollars, and be without the data for a couple of weeks. So I looked for a better solution.

It turns out that Kroll Ontrack, the same company that recovered the data from the hard drive on the doomed Columbia space shuttle, had an office in Reston, Virginia, only a few miles from my office. I got in touch with Kroll Ontrack, set up an appointment with Peter Brown in the Reston office, and drove there with the server in my car.

As I had suspected, the server’s drives were fine. The server’s firmware had indeed failed making it unusable, but with four hard disks that contained useable data. Brown logged the drives from the failed Buffalo server into the Kroll Ontrack recovery system, and then copied the contents to the company servers.

Once he had done that, David Logue, senior lead data recovery engineer for Kroll Ontrack, reconstructed the RAID array as a virtual device. Once that was done, he was able to copy the reconstructed data to an external drive that I picked up at Kroll Ontrack’s Reston office a few days later.

This sounds simple mainly because in my case it wasn’t a complex problem. All of my drives were intact and operational; the data was stored in a standard format for Linux so there were no access problems. But it’s not always that simple.

“We did stuff from [hurricane] Sandy last fall,” said Mike Burmeister, director of Recovery Operations for Kroll Ontrack. “We had hard drives that had been underwater. We’ve had pretty good success.”

Burmeister said that the company was able to recover all of the data from a storage area network in which a disgruntled former employee had gained access and erased all of the data and the backups. After recovering the data, the Kroll engineers were able to help the FBI track down the perpetrator.

But much of the time the data recovery is fairly routine. A server dies, or there’s a fire or a flood and the data becomes unreachable. If companies have backups, then they’re safe. But with the decrease in the cost of storage and the resulting increase in the amount of stored data, backups aren’t always easy. Exacerbating that is the difficulty of storage backups even when they’re available. And not every cloud service has the ability to do fast backups of vast quantities of data.

Fortunately, a professionally run data recovery company such as Kroll Ontrack can get data back from a seemingly unrecoverable place with remarkable success. But there is one thing to remember, as I was reminded by Mike Burmeister when we talked, and that is to make sure you store the recovered data in more than one place. You don’t want to lose it all over again when the next server fails.

Do you know storage? Try our quiz!

Originally published on eWeek.