portable hard drive fails from under use

[esh]

HDD motors go all the time (this will give you an OK power light but no spin-up). Honestly, you can't do statistics on a handful of drives. If you've encountered hundreds (preferably thousands) then you can probably give real structure to what you find. I've had 6 WDC drives go vs 2 Maxtors, but I almost always prefer the WDC ones. I don't think they are less reliable (in fact, more so). Weirdly, if you have a drive that is having trouble spinning up, I have put them in the fridge before and found this has brought two drives (temporarily) back to life. It also worked for one friend of mine too. I am not sure whether this is a fluke yet, but there is a vague physical hand-wavey argument for that it might do *something* at least.

I never bothered looking into online backup services. 1TB+ would not only be costly but take forever on a home connection. I use RAID setups of drives nowadays with one full backup.

[Technical Ben]

The accepted idea is that most faliures are either, platter crashes (not usually fixable) or component failures.
With component failures, it might be a tiny wire or circuit that has cracked. Or a transistor/capacitor overheating and warping it shape, loosing connection. So a quick dump in the freezer will cause metal parts to contract, hopefully regaining contact. Then you have a little while until it starts to bend or crack again.

I do know the cold effects magnets too, but doubt these kind of temperatures are doing anything there.
Google had some good data on how to backup, and how many drives fail.
The usual "have a second hardcopy" as a live backup system, such as raid, can always fail due to a power cut.
http://www.engadget.com/2007/02/18/massive-google-hard-drive-survey-turns-up-very-interesting-thing/
and if you google it, you gets tonsof data to crunch. :F
http://www.google.co.uk/#sclient=psy&hl=en&q=google+hard+drive+reliability&aq=0&aqi=g1g-o1&aql=&oq=&gs_rfai=&pbx=1&fp=8d8451eb556d4e24

Oh, and the wiki article on data backup also suggests, having too many small HDDs is worse than having 2 or 3 large ones. As the more drives you have, the more likely it is you will break one of them. 

[Gary]

Oh yeah Hitachi are another brand to avoid.

Allegedly it was the failure of a Hitachi Data Systems RAID controller that brought a good proportion of Barclays' ATMs plus online banking down not once but twice.

Had an Hitachi in my Voodoo that is still going from what I hear, 5 years and counting Mitch, thing is the old IBM deathstar Deskstar gave Hitachi that tag as they took that brand on just like the bodged firmware on the Seagate Barracuda 7200.11 drives. I think most drives have equal chance to fail or live these days, some batches are bad some good, luck of the draw.

[Rik]

I find that USB bus-powered drives tend to fail more than self-powered.

[Gary]

I find that USB bus-powered drives tend to fail more than self-powered.

I avoid them, Rik. The worse drives I ever had were external Maxtors and a few external WD ones, but that was years back, they all seem pretty much the same now, the Seagate 7200.12 1TB  are only two platters like many big drives these days so density is better.

[Technical Ben]

Strange. Some of those articles were saying Hitachi were the best. But as said. It all depends on setup, external or internal, and which model you get.

[Rik]

I avoid them, Rik. The worse drives I ever had were external Maxtors and a few external WD ones, but that was years back, they all seem pretty much the same now, the Seagate 7200.12 1TB  are only two platters like many big drives these days so density is better.

I like dense, it makes me feel at home.

[Ray]

 

[pctech]

I called a halt to the backup at 2 a.m. this morning as think there was a lot of garbage in there bulking it out.

[Rik]

That's the trouble with modern hard drives, Mitch, space is so cheap that we don't bother to clean out old files.

[pctech]

Going to sit and go through my My Documents folder over the weekend and then give it another go, am on a trial at the minute so not as if its actually cost me anything.

[Rik]

I have documents going back to 1989 - I hoard!

[armadillo]

I called a halt to the backup at 2 a.m. this morning as think there was a lot of garbage in there bulking it out.

I do not quite understand why you need an on-line backup if your data volume is as small as I thought you said. You said the backup was 2.93GB. Maybe that is just a test. But that kind of volume you could backup in 15 minutes onto a pen drive that costs about £5. True it would not give you geographic diversity.

http://www.scan.co.uk/products/4gb-dane-elec-pearl-silver-usb-20-pendrive-retail

That is a USB pen drive 4GB for £5. Why not buy three of them and then you can have three backups, used in rotation so you overwrite the oldest one each time you make a new backup?

Seems like an ideal solution for small volumes of data like a few GB. Not practical for terrabytes though!

I did wonder about farming out some of my external drives to friends for location diversity. Trouble is, my friends are not geeky like me and they would just think I am weird! (Which is probably true).

[Technical Ben]

Or leave it in a bank... no, don't do that, over kill and WAY more expensive.
Hmmm. I can see a business plan here, gap in the market...

[pctech]

I am probably going to use DVDs for now.

[Rik]

Don't trust RWs, Mitch.

[pctech]

TBH thats why I was looking at an online backup solution so it couldn't be lost or wiped by being too near something magnetic.

[Rik]

Like me, you mean.

[pctech]

 

[esh]

Oh, and the wiki article on data backup also suggests, having too many small HDDs is worse than having 2 or 3 large ones. As the more drives you have, the more likely it is you will break one of them. 

Unless it is RAID of course. Excluding RAID-0.

Even without RAID, the question is do you prefer to lose *some* data with a greater likelihood by spreading your backup around, or lose it entirely with less chance by putting it on a single drive?

I never run unraided drives on workstations or backups at all anymore. It's just too dangerous.

[Technical Ben]

TBH thats why I was looking at an online backup solution so it couldn't be lost or wiped by being too near something magnetic.

Been fine since we moved off of tape. Never heard of a HDD dying that way. Unless it's a super magnet.

Oh, and EVEN on raid esh. The core problem is, the more components you have, the more complicated the setup. The more complication, the more likely it is to fail. So if 1 in 10 drives fail, and you have a raid of 10 drives, one will fail. However, if you only have 5 drives, your twice as safe as before. So you keep your data, but are forever replacing dud drives. Maths/statistics is strange like that.

[esh]

Right, I missed your point. Yes that's true. But drives are cheap! Though SSDs are meant to last a few hundred years or so. I'll report back in 2300 and let you know how it's going!

[armadillo]

I inherently mistrust RAID as a method of backup. I absolutely agree with pctech about complexity.

My view on RAID is this. The ideal use for RAID is when you need (at least a fighting chance to maintain) continuous on-line availability of data. An on-line air ticket reservation system, for instance, depends on continuous uptime. So you update a RAID array. Then if a drive fails, the data should still be available on-line until the failed devices can be restored or replaced.

But I do not regard that as a backup policy and I consider it makes a dangerous backup policy. When RAID fails, it often fails catastrophically. TBB lost an entire RAID system a while back. Its purpose is not backup. Its purpose is continuous on-line availability. No other storage method can provide that. But to provide backup, you need data storage that is logically and physically independent of the data being backed up. RAID very obviously does not provide that. Its viability depends on the successful operation of interlinked components. So for ensuring no data loss, it is at least as dangerous as not backing up at all.

I believe a good strategy in an on-line system is to use RAID to provide, or to have the chance to provide, continuous availability of prime data and to use backup to devices that are not part of the RAID array as a means to guard against data loss.

I go so far as to say that RAID is not a backup mechanism at all. If you use RAID, your data is not backed up. It is just provided in a package to maximise the chance of continuous availability if one of the drives fails.

That is my opinion, anyway!

[pctech]

RAID was developed before I cared much about computers but as 'dillo says it is my understanding that its primary reason for development was to provide a mechanism for orderly automatic failover in the event of a disk device failing in a server/disk array, however due to the constant replication required to maintain a perfect mirror image of the primary disk all disk device in the array are placed under the same duty (OS writes to primary disk and RAID writes a copy to each other disk in the array) so the wear and tear is likely to be equal on a heavy duty system so it does not negate the need for a backup.

In a lot of organisations including my own employer RAID maybe used in the front end servers to provide OS uptime but business data is often stored on large Storage Area Network devices which use their own proprietary OS and file systems but which are also regularly backed up.

[armadillo]

Oops. Glad you agree Mitch. I also apologise to Technical Ben for mis-attributing his comment on complexity. Sorry Ben!