[Matthew Attoe]

Hi,

For years, whenever I've bought a new hard disc, I've always erased/formatted it by writing zeroes to it before using it? My - probably flawed - logic is that any bad sectors on the drive will be flagged by the writing zeroes process, hopefully making the drive more secure/stable.

Is this worth doing? Or is it a waste of time?

Many thanks,

Matthew

[Simon]

Personally, I think it's a waste of time. Of course if you have the time to waste (over night for example) and it makes you feel better you can continue doing it.

[shifuimam]

It IS a good idea to zero out a hard drive if you're going to sell it (or the computer it's in) on eBay. It's too easy to recover data from formatted hard drives. Zeroing it out makes it almost impossible to recover anything.

For your own drives, however, it's unnecessary.

[Simon]

Originally Posted by shifuimam 

It IS a good idea to zero out a hard drive if you're going to sell it (or the computer it's in) on eBay. It's too easy to recover data from formatted hard drives. Zeroing it out makes it almost impossible to recover anything.

The OP is talking about new disks. He wants to know if he should zero when he buys a brand new drive. He never even mentioned selling an old drive.

[ghporter]

Brand new drive? Not necessary. Any bad blocks or sectors would have been flagged by the factory in the low-level formatting, and writing zeros to the whole thing would simply take up some (probably a LOT) of your time.

Disposing of a drive? VERY necessary. Not that your prized collection of rare digital copies of TMNT drafts would fall into the wrong hands, but that ANY data left on the drive could be used against you. Shifuimam is spot on, because there are free tools that let you recover as much as 100% of the data from a "formatted" drive-I've used some to recover "ooops!!!" systems.

Caveat: I'd use at least the 7 pass option in Disk Utility because while significantly "non-trivial," it is possible to recover data that's been overwritten once with just zeros, while multiple passes with a variety of characters (specific hex characters like "f0f0" then "0f0f", then random characters, etc.) make this far more challenging, and 7 passes makes it virtually (note the hedge here...) impossible to recover data without really serious (and expensive and time consuming) laboratory processes that may not work anyway.

[turtle777]

I thought that todays HD automatically re-map bad blocks on the fly when they are encountered.
They set aside some extra space specifically for that purpose.

-t

[OreoCookie]

Regarding writing zeros once or many times: very recently, a group of security experts were testing whether it's necessary to write zeros once or several times. Writing zeros once is more than sufficient: the probability to get one bit right were 56 % -- you get 50 % for free by just guessing! To probability to recover a single byte correctly is already below 1 %. I won't bore you with the math to recover usable amounts of data.

@Turtle
Yes, that's what they're supposed to do. You can also initiate a surface scan via smartmontools at any time you like (e. g. after you've dropped your mobile Mac).

[turtle777]

Originally Posted by OreoCookie 

Regarding writing zeros once or many times: very recently, a group of security experts were testing whether it's necessary to write zeros once or several times. Writing zeros once is more than sufficient

Thanks for the link.

It goes right along my thinking that Gutman's claims have always been overhyped.

-t

[ghporter]

Originally Posted by turtle777 

I thought that todays HD automatically re-map bad blocks on the fly when they are encountered.
They set aside some extra space specifically for that purpose.

-t

Precisely. So it's all taken care of ahead of time or so transparently that you never see it.

Originally Posted by OreoCookie 

Writing zeros once is more than sufficient: the probability to get one bit right were 56 % -- you get 50 % for free by just guessing! To probability to recover a single byte correctly is already below 1 %. I won't bore you with the math to recover usable amounts of data.

I have a long history of digital security paranoia (it was my job for quite a lot of time), so i always suggest going that extra mile. Of course a 7-pass wipe will take at least 7 times as long as a one-pass wipe, so balance the "paranoia" factor with how much you value your time...

Originally Posted by turtle777 

Thanks for the link.

It goes right along my thinking that Gutman's claims have always been overhyped.

-t

Gutman was a scary optimist. He liked to take theoretical possibilities (in wonderfully perfect, invented scenarios) and scare people by VERY optimistically suggesting that he could regenerate data. To my knowledge he never provided a timeframe for how long it would be before he could do that, nor how much hardware and computing power it would take. That's a big sign right there...