Longevity of hard drive data

bw wrote on 8/10/2006, 5:15 AM
In a recent post 'jrazz' mentioned the practice of archiving to HDD. With the current price of drives this is a very effective method. However what is the life of data on a drive which is just sitting there on the shelf? Surely the disks will eventually demagnetise? If so is there now a program to "refresh" the stored data? (takes you back to core memory which "refreshed" after every read). Or am I worrying about nothing?


farss wrote on 8/10/2006, 5:34 AM
Well years ago I used to use core memory. I remember loading code into, putting it back into the store for a long time and still the code was there so I don't think ther data on a HDD would fade away, I'm at the moment transferring 1/4" tape recorded decade ago and the 'data' is still there and binary data is much, much harder to erase.

However disks are more than just data. I've had one that was fine when put on the shelf but lasted only a few hours when put back into service.

Conversely I've got PC with HDDs that have been merrily spinning away for over a decade so my gut feeling is they're more reliable if left running. Certainly a RAID 5 NAS box would seem the most secure, the cost of the box / controller adds only a small amount to the cost of the drives if using the larger units. You'd need to factor in the cost of power over all the years as well I guess.

Chienworks wrote on 8/10/2006, 6:04 AM
It is true that drives left sitting for a long time can develop what's called "stiction". Various mechanical parts will begin to freeze up as lubricants settle. I'm sure that given enough time metal-to-metal joins will begin auto-welding too, though i imagine that would take longer than decades.

Note ... this is *NOT* a recommended data recovery method!!!!! But, in an emergency, i've had good success with stuck drives by simply giving the drive a sharp rap on the table top or workbench before mounting it back in a PC. I'm sure this does more damage than good in the long run, but in the short run, it will probably keep an old ailing drive running long enough to copy all the data to a new drive and then some.

Given the choice between running a drive live for 10 years just to preserve data or leaving it sitting on a dusty shelf unpowered, i would definitely choose unpowered. Very little can go bad with an unpowered drive short of earthquakes, fire, flood, meteor strike, gunshot, EMP*, etc. which would all affect a live drive in a PC just as badly. There are so many more things that can endanger the data while a drive is running that it's almost inevitable that something bad will happen to a powered drive over the course of 10 years or more. Live drives can suffer hacking, viruses, head wear, surface wear, overheating, bearing wear, lightning and power spikes, malicious (or unintentional) data tampering, and so on. Compared to all this, dealing with stiction seems rather trivial.

*If your workspace suffers one of these ... why the heck would you be worried about your videos?
farss wrote on 8/10/2006, 6:52 AM
Well I've gotta disagree about leaving drive on shelves, just my opinion that's been partially swayed by having at least one go bad within 18 months after doing just that.

Firstly the head and spindle bearings are pretty delicate so they could well suffer the auto-welding you mention much quicker, that's certainly what seems to have happened in my case, yes it came unstuck, ran for a few hours (long enough to install Win2k)and then lost vital OS files. I've since run Checkdisk which found a number of very large blocks of corrpupted files, sounds like a head crash. Only a 80GB drive so no great loss.

But also, yes modern electronics (as opposed to electromechanics)left under ideal storage conditions should fair better than if it's powered up but I've seen some pretty bad cases of what happens to electronics under less than ideal storage conditions. One of my colleagues is now trashing a huge amount of once working vintage video gear. Basically left in a warehouse for a long time and corrosion has eaten into everything and the dissimilar metals on PCBs corrode very easily. With the power on things stay warm so less risk of condensation.

My suggestion to anyone who doesn't want to leave drives powered up, seal them in an airtight storage container (Tupperware or ziplock bags are cheap) with a packet of silica gel.

I've even seen electronics that was powered up fail within a few years from corrosion, relay coils turned to green blobs, SCRs go leaky from corroded metal / porcelain seals. About 2 KMs from the ocean and 20M from the battery room, which was causing so much corrosion I don't know. Thing is HDDs are not airtight, so any gaseous pollutants can get inside.

johnmeyer wrote on 8/10/2006, 7:09 AM
However what is the life of data on a drive which is just sitting there on the shelf? Surely the disks will eventually demagnetise?

If properly stored, the magnetism will almost certainly outlast you. I have audio tapes recorded in the early 1950s that play back just fine.

The mechanical and electrical issues are a bigger concern. I have a laptop I purchased in 1987 that runs 24 hours a day, but with the disk "spun down." I use it to record phone calls from the SMDR output from my PBX. It records to floppy. About once every two months, I have to do maintenance, and I spin up the (20 MB) internal hard drive. So far, it still works fine. Hardly a scientific study, but it shows that older hard disks can still work. I also have a 486 computer that sits under the stairs that I bring out every few years to amuse my now almost-grown children to show them what they used when they were small. It has a 2 GB hard drive that spins up just fine. I think I purchased that around 1993.

The electrical drive electronics might go bad, although the main problem with electronics has been electrolytic capacitors, and that has been mostly with those used in switching power supplies, and the problem was usually associated with supplies turned on 24/7 (like fax machines, VCRs, etc.).

Bottom line is that drives are probably not the greatest archival medium, but should be perfectly fine for 5-10 years. For archiving, if you get archival DVD (you can search these posts for brand names), I am quite certain that they will last a LONG time (50-200 years) if they are stored properly (no light, and room temperature or slightly less). I am also pretty certain that you will be able to get equipment that reads DVD or CD for many decades to come, thus ensuring that you or your heirs can transfer the data to some other media, when that time comes. (I say this because 3.5" floppies were introduced in 1984 and CD-ROM a year later, and they are still both being manufactured twenty years later in an industry where many things come and go in less than five. It will take a LONG time for all this stuff to totally disappear and go bad. My first Sony 1x SCSI CD-ROM still works, as do two 3.5 floppy drives purchased in the mid 1980s.)
RalphM wrote on 8/10/2006, 8:03 AM
One technique I have used with a drive that refused to spin up was to put it in the freezer for a few hours, then power it up. It worked, and I'm assuming that the temperature change caused dissimilar expansion sufficient to break the "sticktion".

Do this only in a low humidity envioronment and assume that you have only one chance to get the data you want....

Worked for me, but could have also been a disaster, so try at your own risk.
Jayster wrote on 8/10/2006, 8:25 AM
That does indeed sound quite risky! I assume you put it in a ziplock bag to avoid condensation (but can it really be avoided?). And the freezer would cause contraction, not expansion, of the metal parts.

As you suggested, it sounds like you should plan to throw away the drive after that.

One time I dropped my cell phone into the water at a water park, and of course it quit working. I put it in the oven at low temperature to dry it out. That made it able to power on but the battery couldn't last. A few weeks later it was working fine.

As they say, "don't try this at home" or something like that...