Three Things are Certain… A Cautionary Tale

12Dec11

Originally posted on www.hmml.org 11-7-2011

From a humorous series of “Computer Haikus”

Three things are certain,
Death, taxes and lost data.
Guess which has occurred?

Remember when a floppy diskette held lots of data? Data was fairly simple back then: databases, accounting spreadsheets, word processing files, mostly text and numbers. Most of this data was the stuff of professionals, the people who worked with computers as part of their jobs. Today, people generate immense quantities of digital data as part of their everyday lives. Text and numbers are still part of our digital realm, but we’ve added digital video, photographs, and music. Storing and safeguarding all this data is something that most folks don’t pay much attention to. Mostly, things seem to “just work.” When they don’t, it can be a real problem.

HMML’s manuscript digitization projects likewise generate huge amounts of data. Right now, we have 167 individual external hard disk drives containing manuscript image data, totalling about 115 terabytes (a terabyte is 1024 gigabytes). From the perspective of 20 years ago, this is an unimaginable amount of data, and making sure that all this data is safe, backed up, and accessible is a large part of my job. We do everything we can to avoid problems, but, as the haiku says, lost data will occur at some time.

This happened to HMML with a disk drive from Jerusalem containing the files from our latest digitization project. It was a typical external hard drive, the kind one can get in any electronics store. HMML has dozens of these, and they’ve been a reliable and economical way to transport and store large amounts of data. The drive worked perfectly when it arrived—I had “mounted” the drive on both PC and Mac computers a number of times to catalog its contents. The manuscripts on the drive had been viewed by several colleagues without any problems. Usually, they would have looked at the images on HMML’s file server, but I was waiting for more disk space to be allocated on the server and had not yet uploaded the drive’s contents.

Then, it stopped working. One day, I plugged the USB cable into the drive, then the power cable. The tiny LED light, which usually blinks while the drive is mounting up, simply blinked once, and there was no sound of the motor spinning up. Then nothing, just a black plastic object with several months of work on it that wouldn’t spin up or mount on the host computer. I tried it on a different computer with the same result.

hard drive

Typical External Hard Drive. Simply a black plastic enclosure with ports for DC power and USB data cable.

What to do? My first thought was that the drive itself (the metal thing buried within the exterior housing) was probably alright, but that the bridge assembly containing the USB and power connections had failed. To test this theory, I figured I could remove the “bare drive” from the enclosure and put it into an empty drive enclosure (these can be purchased from computer supply shops) to test it. We happened to have one of these enclosures at HMML, so this seemed the logical first step.

Drive opened

Inside the Plastic Box. "Bare" drive fitted in enclosure with rubber mounts. USB/power bridge is on left.

Getting into the drive was interesting. This particular model is a plain plastic box with no fasteners. Flipping the thing over revealed a perimeter seam that one could get the blade of a knife into. By carefully prying around the bottom, I could eventually remove it, revealing the bare drive held in place by rubber corner pieces. There were almost no fasteners of any kind. After this, it was an easy process to remove the bridge and place the bare drive into the new enclosure. All went as planned, but when the new assembly was connected and plugged in, the result was the same, no spin-up.

Next, I turned my attention to another potential failure point on the drive: the motherboard. Yes, a hard disk has its very own motherboard that sits on the outside of the drive, communicating with the metal platter and data-reading heads that are inside the drum-like enclosure. This small green board is attached to the hard disk with special fasteners and contains dozens of small microchips, spring-loaded electrical contacts, and the connector that the USB/power bridge attaches to. Perhaps something on the motherboard had failed, but to test this, I would need a replacement.

Back of drive

Drive Motherboard. This where I think the failure happened. To fix this, an identical replacement board would need to be obtained.

Amazon.com to the rescue! I was able to order a new hard drive with the same model number and have it arrive in a couple of days. The motherboard looked identical to the one on the broken drive, so I swapped the boards, put the bridge back on, and reassembled the unit. Then, the moment of truth as I plugged it in and fired it up.

Bare drive

Bare Drive. This is what arrived from Amazon.

Progress, but not success. This time, I could hear the motor spin up and continue to run, but the drive wouldn’t show up on the computer. I then checked the labels on the two drives. The main model number was the same, but the long, secondary model designation was slightly different. I guessed that there was probably some minor firmware difference in the two motherboards, just enough so that although power could be delivered to the defective drive’s motor, it couldn’t interpret the data on the platter itself.

Defeated, I looked at the option of sending the drive to a professional firm that specializes in data recovery from broken or damaged hard disks. They said that they probably could get the data off the drive for between $600 and $1000. Ouch!

During all this experimentation, I wasn’t overly worried because of one thing—I had a backup copy of the data. Right from the start, I contacted our field director in Jerusalem and had him copy the manuscript images to a new hard drive and ship it to HMML. It arrived in about a week and functions perfectly.

The lessons learned? For starters, one has to have a backup strategy in place before something goes wrong. At HMML, each project location retains a complete set of the manuscript images produced. Once HMML’s copy arrives, more backups are made, totaling five in all.

Having multiple copies isn’t enough; the backups need to be in different locations as well. That way, if some physical disaster (fire, flooding, etc.) destroys data in one place, the backups are in a different, safe place. The manuscript data for HMML’s projects ends up in five different places, one going to a special storage facility in Utah.

People should adopt a similar strategy with their personal data. My own data resides on CD and DVD disks in one place and external disk drives in another. At some time, the data will also be copied to one of those small “pocket” hard drives which can fit into a bank safe deposit box.

Keeping track of all this and taking the time to make and safeguard the copies is a bit of work, but consider the consequences of having all of one’s data disappearing in an instant. I hear about this from time to time: someone’s computer stops working and takes with it the person’s financial files, email history, music, pictures, everything…

Don’t be that person. Hard drives are inexpensive enough these days for anyone to put together a good data backup plan. Get organized, take some time in making sure all your data is copied, and figure out a number of safe places to put the backups. Then, you can go back to simply worrying about death and taxes.

Advertisements


No Responses Yet to “Three Things are Certain… A Cautionary Tale”

  1. Leave a Comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: