A Walk Through the Hard Drive Recovery Process

Fields writes "It's well known that failed hard drives can be recovered, but few people actually use a recovery service because they're expensive and not always successful. Even fewer people ever get any insights into the process, as recovery companies are secretive about their methods and rarely reveal any more information that is necessary for billing. Geek.com has an article walking through a drive recovery handled by DriveSavers. The recovery team did not give away many secrets, but they did reveal a number of insights into the process. From the article, "'[M]y drive failed in about every way you can imagine. It had electro-mechanical failure resulting in severe media damage. Seagate considered it dead, but I didn't give up. It's actually pretty amazing that they were able to recover nearly all of the data. Of course, they had to do some rebuilding, but that's what you expect when you send it to the ER for hard drives.'" Be sure to visit the Museum of Disk-asters, too.

Their secret revealed...

[Anonymous Coward]

A hard drive shaped freezer.

Summary

[Anonymous Coward]

"It's well known that failed hard drives can be recovered"

[Citation Needed]

Nice freaking advertisement

[meeotch]

"Although there was severe media corruption on this drive, DriveSavers engineers were able to successfully recover the majority of the critical data by utilizing our proprietary software and methodology."

I'm sorry, but that was the most content-free load I've read on /. in a while. And no, I'm not new here - I just usually don't RTFA. ;-P

Re:Their secret revealed...

[Anonymous Coward]

That only works if you have "engineers wearing specialized, certified cleanroom garments" that place the drive in the freezer. And don't forget to charge $1500.

Re:How do you backup

[mrbluze]

Google cache?

Re:Summary of Article

[cojsl]

you forgot:

4. profit!

Re:Hmmm. Cheaper... well, I s'pose the depends

[davidsyes]

on where you are, and that determines how hard condoms are to come by... (Careful... I'm watching my prepositions, and not necessarily the pre-positions....)...

But, in the vein of hard drives, i guess a not having protection can cause a bunch of thrashing about in the end... (buns, umm, PUNS intended...)

Re:Never had any luck with recovery

[syousef]

I had a drive with a mechanical problem that wouldn't spin up. It didn't have anything critical but it did have my last x weeks worth of software downloads which would have been a pain to re-download. I tried banging it, freezing it, you name it. What worked in the end was making sure it was upside down when it was powered up and giving it a little tap to get it spinning. Got it running for 24 hours - long enough to get all my data off. About 200Gb. Obviously a mechanical failure and obviously pure luck that I got it working again.

No, I'm New Here

[New Here]

No, I'm New Here

Re:This may be a dumb question...

[bonhomme_de_neige]

how easy/hard (compared to magnetic HDDs) is it to permanently blast data off a flash drive if you don't want the data found?

Much easier - 10 minutes with a mortar and pestle pretty much guarantees recovery will be impossible. That method would take a lot longer (and require more equipment) for hard drives.

Assuming, of course, that if hiding the data is that important, the cost of a flash drive is a sacrifice you're willing to make. ;)

Re:Never had any luck with recovery

[mpaulsen]

"You might try reading the article linked in the parent posting for a case where the drive had both bad electronics *and* the head crashed yet they still recovered 80% of the data."

80%? People get paid for this?
Guessing 1 or 0 for every bit will successfully recover 50% of the data, assuming the ones and zeros are equally represented.

Once you've got it 50% recovered it's a simple matter to flip the bits in the remaining 50%. 100% recovery.

Re:Hmmm.

[mortonda]

Ah young love. ;)

Yes, once a geek discovers the beauty of a good backup system, he/she has stepped into a new world.

My backup/archive server is my most lovingly maintained system. It has saved me several times, and recently had to go through a hard drive replacement. That had me nervous.

Re:Advertisement warning...

[NormalVisual]

I don't think Class 100 really qualifies as "ultra-cool" in the clean-room world, but it *does* however have a certain cachet of inconvenience as far as having to take a non-trivial amount of time to get into the bunny suit, walk across the 50 feet of sticky mats, through the air showers and into the actual clean-room area only to discover *then* that you have to take a leak. I've had it happen more than once.

Re:DriveSavers

[Anonymous Coward]

Whoosh!!!

Re:Never had any luck with recovery

[mpaulsen]

I think you're math is off... 1s and 0s aren't data in any way, shape, or form. There is no useful "data" at the user level stored in 1s and 0s. Data is chunks of 1s and 0s that make up stored files that are actually useful to the user"

You're mostly correct. Individual ones and zeros are called anecdotes. It's only when you put together two or three anecdotes that you have data.

so having 50% of the file uncorrupted is not a possibility. Corruption is all or none, one bit is wrong and there is no data

You're missing the beauty of the algorithm. You simply take a guess at each bit. If you're right, you've recovered that anecdote. The anecdotes are binary, so if you guess wrong all you need to do is flip the bit.

the idea of partial corruption is illegitimate for all intents of purposes

Missed it by -| |- that much. So close....so close.

because any amount of corruption is the same, save for the fact "less" corruption may make recovery easier.

I could try to explain the theory, but it would be easier if you just tried it yourself. Start with 10101 as your data and corrupt it any way you want. Now flip a coin for each bit and record a 1 if you flip heads and 0 if you flip tails. Keep the bits which are correct and flip the ones which are incorrect. You just recovered 100% of the data.

Re:Never had any luck with recovery

[mpaulsen]

Hmm.. okay I think I understand now, but then how to you know what bits are correct? Isn't that what you are looking for?

Just diff against your backups.

Re:Nice Theory But...

[KlaymenDK]

Fine, so it's the air leaking out and ... hardening ... yeah.

Re:Nice Theory But...

[chihowa]

Air is a fluid.

Re:Their secret revealed...

[Sobrique]

Why bother, when /dev/null is so handy and quick?

Re:Nice Theory But...

[Jurily]

Back in the days when I did a lot of Win98 installs, I always carried a nice big hammer around. Put it on the computer before booting it up, and it installed perfectly *every time*.

It started as a joke, but it actually worked... guess those casings had built-in hammer detectors, and the installer checked for it.

Re:Their secret revealed...

[ajs318]

Because /dev/null can get full, if you put too many ones into it! Then you have to dd /dev/zero into it to balance it out. The zeros and ones end up annihilating one another and the cosmic balance is restored.

Re:Nice Theory But...

[Reziac]

Kinda like my workbench HD, an ancient (800 mb) W.D. I got it as a failed HD from a customer, and it was then about 5 years old. I gave it the "beat it ever-harder, until it either gives in or dies" treatment, and....

Powered on... just a whir, no bootup:

rrrrrr... [tap tap]

RRRRRR... [TAP TAP]

*R*R*R*R*R*R*R... [*!!*WHAP*!!*]

And after that it was unstictioned for good, and has worked fine ever since. It's now 13 years old!!