I found this interesting, it compares Hitachi, Western Digital and Seagate drives for long-term reliability. The test was done using thousands of drives in 24/7 real-world server conditions.
Under no circumstances should anyone rely on anything in the "cloud" for preserving digital content for decades. Companies come and go, and policies can get changed at the stroke of a pen.
Cloud storage is very useful for intermediate-term backup, but for archival storage, you need to do it yourself.
My business is preservation and restoration of various types of media, both analog and digital. Some types of media are almost ageless: 100-year-old 78 rpm records are as good as the day they were recorded, if they haven't been played. Audio tape from 1950 (which I've transferred) sounds as good as the day it was recorded, although some tape formulations, depending on storage conditions, can shed, sometimes to the point where the tape is no longer any good.
B&W film is absolutely amazing, and I have transferred negatives which are over 100 years old that look as good as the day the picture was taken. Color film is a mixed bag, and most of it fades quickly, but Kodachrome is absolutely amazing: I have some 1942 Kodachrome that has not changed at all.
Rotating magnetic media has so far generally been very reliable. I have no problem with 5 1/4 and 3 1/2 inch floppies from the 1980s (I don't have any equipment to read 8 inch floppies). Zip disks were never any good, and they have proven to be a problem even after only two decades. Hard drives have generally stored well, but the mechanics sometimes seize up. I have had problems with MFM (mid-1980s technology) drives. The problem has nothing to do with the interface (MFM) but instead with the bearings and motor. A clean room disk drive recovery company could probably recover the information, but that would be thousands of dollars.
Optical media so far has been perfect, but only with good media. I have had "bargain" media DVDs fade away after only a few years. Some laserdiscs quickly developed "laser rot," although that was a manufacturing defect that usually showed up within a few months, not years.
With quality optical media, it appears that extreme longevity is very likely: my Taiyo-Yuden DVDs still test the same now as when first burned many years ago, and accelerated aging tests seem to indicate that these discs should last for many decades, and perhaps a century or more.
So, from everything I've read, and from my own experience, optical media (DVDs in particular) seem to be the best option for long-term (decades or more) archiving. As for being able to read optical media, I think there is much, much higher probability that this standard will propagate for decades into the future, unlike disk drives where the drive interface disappears quickly (IDE is starting to disappear). The drives for optical media are like phonographs that could still play 78 rpm records for decades after the last 78 was issued. The analogy is that current "round shiny disc" players (DVD and Blu-Ray) can still read CD discs thirty years after that media (in audio form) was first made available to consumers.
Yes, it has been thirty years since the audio CD first hit the market.
Finally, the jury is definitely still out on using flash media for long-term storage, and until someone can find an accelerated aging study, I would recommend against relying on this media to store anything for longer than a decade.
If you are concerned about your Seagate drives, run a SMART drive test (you can download some free software). It can give you an idea of whether a problem is brewing.
One VERY significant advantage that rotating disk drives have over the new SSD technology is that they usually fail gradually and "gracefully." If you use monitoring software, and even if you don't, you can often recover most -- and sometimes all -- data from a failing rotating magnetic media drive. By contrast, SSD drives apparently (I have no direct experience) fail all at once, without warning.
I use SSD drives in a cache configuration on my new laptop, and this is a great use of this technology because reading from SSD is much faster than hard drives, but writing is often slower. According to the cache software, about 90% of all reads are now from the SSD. Thus, I get the speed advantage without having to worry about any catastrophic failure, and without having to worry about write cycle failures.
I am not yet ready to rely only on SSD for my primary drive.
This article is interesting from the sense that I have had the exact opposite experience. Two IBM drives I had failed (never bought them again) and one WD and one Hitachi. I currently have been running some Seagates for years now approx. 8 with no problems. At one point Seagate goobled up Conner and I wonder what the tech impact was on reliability?
I am transitioning to SSD now and they are supposed to be good for 10 years (or 40 in the case of the memory type in the Samsung 840 Pro). I am very curious as to how they hold up.
So, to me, it looks like the only way to read this stuff in 50 years is like it was 50 years ago: don't throw out the old Nagra deck and keep the 1930's reel to reel machine in pristine condition because you might need it later.
Oops, there was no reel-to-reel tape around in the 1930s.
My drives are Samsung 840 Pros, as they have one of the best reputations for reliability. The worst was OCZ, absolutely appalling devices.
But I have always regarded spinning disk drives as consumables, and if they were still around at 5 years, I used to habitually replace them. I used to build large capacity hard drives for use in aircraft on-board systems, and the drive of choice back then was always IBM.
Ehemaliger User
schrieb am 24.01.2014 um 02:25 Uhr
I think the moral is contained in the report we are discussing. Don't be afraid to use any reasonably priced and reasonably reliable disk, PROVIDED THAT YOU MONITOR ITS PERFORMANCE. You can't just install it and forget.
Install software such as Hard Disk Sentinel that will let you know when reallocations are occurring. The RAID software on my computer gives a report during BIOS boot time: Green = don't worry, Yellow = watch out, Red = you are in trouble.
SMART isn't very reliable. Last weekend I was with a guy who had his hard drive fail right there and SMART never said anything was wrong. It's never detected a drive going when I've had them go.
I can normally tell when a drive is going. Symptoms are almost always the same: machine locks up frequently for seemingly no reason, files tend to exist then not exist but are still listed, the more stuff you put on there the more issues you have.
JohnMeyer, I installed a Plextor SSD in my system, and they recommended using their utility called "Plextool" which, as you can see, has a graphical readout supposedly showing gradual SSD deterioration. I've never used it so I don't know how it works.
Another nice feature of the Plextor SSDs is that they can be used on XP systems also. XP doesn't implement the Trim command, but Plextor has a built-in "garbage collection" feature which cleans up the disk during idle moments. I've found the SSD to be about twice as fast as the 10,000rpm Raptor it replaced.
Good post.
Just adding my experience over the last 10-15 years and quite different results.
Hitachi being the worst in terms of reliability, all with mechanical spin failures. Lost count how many. Seconded by Seagate with their noisy bearings leading also to failures. Last and best Western Digital. Never had a single failure. Ever.
WDs are the only drive i recommend to anyone simply from past experience.
The only time I ever completely lost data was with Seagate. I was only keeping a single backup, both hard drives were Seagate and both completely died without warning. I now keep triple backups and won't touch Seagate with a barge pole. I do have some Western Digital drives but the most reliable for me have been Hitachi.
i haven't had any problems with seagate drives over many years - apart from the well documented one a few years back, and seagate replaced that drive immediately.
Everyone puts so much confidence in a published report that takes a small sample and makes sweeping generalizations about HDD longevity. Then everyone adds their own anecdotal input.
Here's mine:
MTBF is the MEAN time before failure. Half the buyers will see failure before the MTBF, and half will surpass that. I almost always buy Seagate Barracuda drives. In the past ten years, I have bought about three dozen. Two failed early and Seagate replaced both quite quickly. One is in my "office" PC (the one I use for email and general office duties) it is about three years of intense usage and starting to show signs of pending failure. Time for an SSD there. The editing PC has an SSD boot drive and one Seagate for Windows data folders (My Documents, My Pictures, and user files).
All of my projects are done on external drive docking stations, and when I need to revisit an old project, there has never been a read problem. When I get time I plan to move my archive IDE drives to a couple of 4Tb helium drives.
"MTBF is the MEAN time before failure. Half the buyers will see failure before the MTBF, and half will surpass that"
It doesn't change the general thrust of the statement that disk life is a random variable and that in a small sample, an individual's experience may differ significantly from the general trend, but my pedantic point is that if the M in MTBF stood for "median" then the rest of the statement would be true.
Mean and median will be the same for a symmetrical distribution, but since life expectancy ranges from zero at one end (not minus infinity) and by plus infinity at the other, the distribution will not be symmetrical. The mean will be greater than the median, which may be why manufacturers prefer that statistic. (For example, if one disk never died, its life would be infinite and the mean of all disks would also be infinite, but the median may not change at all.)
The logarithm of life expectancy would be more symmetrical, however, since it ranges from minus to plus infinity.
My experience is similar to Paul W's. While I can't say a WD drive has never failed on me, I can say they last longer (at least in my experience) than any other drives. Worst being Seagate and followed closely with Maxtor (now history). I have only WD Black (none of the 'cheap' ones like Greens) with some now approaching 6 years @ 24x7. I keep them cool though... WD's do not like heat.
I usually upgrade out of a drive before it crashes though. For example, I probably have 2 or 3 80gb drives around that still work but why bother? Last drive bought was a 4TB WD Black about 1.5 years ago and no problems at all.
Like someone else noted, when a drive starts getting slow to access I usually have enough time to limp it along to get the data off of it.