hard disks at archival media

[Doug Lassiter]

I'm familiar with the lifetime of powered up hard disks. Mainly bearings and mechanical stuff, one can usually count on four or five years. But what's the lifetime of a rarely-powered-up hard disk? As in, one that I fire up every few months just to dump some archival material on? Presumably it's less a question of bearing lifetime, and more a question of magnetic domain lifetime. I'd think it would be quite long. If the latter, then the archival strategy would be just to rewrite the disk occasionally. A broader question is .. for the purpose of archival storage, what's the proper strategy for a hard disk?

I'm using M-disks for really important stuff, but on a GB/$ basis, those are more expensive than a hard disk.

 

[Shaun o]

There is really no stratagem, be it based on the wearing of the bearings of the drive or how long the disk platter hold it`s magnetic storage integrity.

Like with most things the more you use it and the time it is operational its life expectancy is reduced as time passes.
If it was in operation 24/7 in a powered up state then it could last the whole 5 years plus.

Or it could simply fail after 2 years.

The same applies to if the drive was used and powered up for short data transfers or copying data to and from the drive in question.

Because you have three factors, the mechanical, the magnetic, and the power, if any out of the three fail in some way since every HD requires each.

It`s hard to say if it`s best to have a drive powered 24/7 or just when needed.

For example switching power on and off can damage a drive if one of the electrical components used to provide power becomes weak.

By doing that act of just applying power to the device it may break one day.
And it often happens.
Because powering up and down creates stress on electronic components.

 

[13thmonkey]

ensure that these are not your only backup/archive, at least 2, and if you are only dumping stuff to them every month or so, then 1 should be offsite (it should be anyway but it's difficult on a daily basis)

 

[Doug Lassiter]

OK, I'm not sure I got an answer. Clearly turning the drive on and off is not going to help the lifetime. But it's not going to be turned on and off a lot. Maybe once a month? So after 5 years, power gets cycled 60 times. You're saying that's a large number?

Obviously this isn't my only archive. Let's say I have two or three drives in different locations. My question isn't about general archiving strategy, but the lifetime strategy for each drive.

There must be a clear answer to this somewhere. What if it was written once, turned off and NEVER turned back on. How long would the data last? Five years? A hundred years? How do on/off cycles compare with drive hours in determining MTBF? How many on/off cycles count for a year of continuous operation in determining lifetime?

I have to assume that this information is available from a reputable manufacturer. Any business spending a load of money on drives for archiving is going to want answers to these questions in mapping out a usage strategy.

 

[13thmonkey]

I think our point was that regardless of the theoretical minimum amount of time before the drive became unreadable, there is always the opportunity for a failure at the next turn on, even if it is only the 100th time being turned on, so the theoretical number is a bit of a moot point TBH.

However given that you have minimised those risks through multiple drives etc. the magnetic domain problem is then your next biggest concern as it is a natural feature of the media being used.

I've asked a WD rep to stop by and see what they think. It could be nominally infinite as it's not being interacted with to lose magnetic energy, barring natural radiation causing issues.

 

[Doug Lassiter]

That's what I was wondering. Old fashioned magnetic tapes are readable (with the right hardware) fifty or sixty years after they were written, but of course the bit density on those was a lot lower.

Actually, now that I do a little research, I head that "bit rot" on magnetic disk medium can be significant, and that hard drives are not recommended for, nor designed for, long-term storage. Actually, businesses that do long term archiving almost always do it on magnetic tape, the lifetime of which is not only long but extremely well studied. One recommendation is to refresh the write on a hard disk every two years.

Of course a rarely spun up hard drive can have it's own problems, as lubricants settle out and become unavailable to the hubs.

 

[Doug Lassiter]

Here's a nice article.
http://www.pcworld.com/article/2984597/storage/hard-core-data-preservation-the-best-media-and-methods-for-archiving-your-data.html

One strategy is disk refresh software, which reads and rewrites a disk. There is a free app for PCs called "Disk Fresh" that will do that.

For a Mac, it's also easy.
https://larryjordan.com/articles/technique-refreshing-hard-disk-storage/

 

[Doug Lassiter]

Well, on thinking about it, I have to be suspicious about the Larry Jordan strategy. The implication is that copying hard disk data automatically "refreshes" the magnetic bits. This sounds a bit like total bunk.

Can anybody here provide any insight?

This is NOT about detecting bad sectors and repairing them. This is about preventing them. I'm kind of amazed that there is so much conversation about this on the forums without any credible solution other than just copying all the data out and back.