Formatting a HD to extend its life?

[Elixer]

I got a mail the other day from a buddy that said one of his clients wants to have him come over and backup and reformat all their storage HDs, and then restore the data.

When quizzed on why you would do that, the client responded that he read that you should always format storage HDs to prevent wearing out the HD from reading the same location and fix any errors.

He then told them this was really unnecessary, and will not extend the life or prevent any errors. Client insisted. So, now he is over there doing what the client wanted at a cost of $2500. (30 machines)

Has anyone else ever heard of anyone else doing such a thing ?
I find it totally unnecessary, and a huge waste of $$$.
While formatting does force a remap of bad clusters, all that does is hide the problem. Most business places that I know, toss the drives once they start getting bad clusters.
As for extending the life of a HD, I also say bollocks. It just adds more work for the HD.

 

[GlacierFreeze]

For $2500, I wouldn't argue with them. lol

There's an idiot, I mean sucker born every minute.

 

[KingFatty]

What about something, ah I can't remember the exact phrase, but it was called bit rot or something? I could see how re-writing the bits, as in re-magnetizing the platter with a fresh run over the platters, could make the magnetic signal stronger, compared to bits that were magnetized many years earlier?

Or is that always happening and bit rot is only when you disconnect and power off the drive?

 

[Binky]

The (paying) customer is always right.

 

[Gunbuster]

How big are their "storage" drives? Honestly $83 per machine seems low for that much scut work.

Can you get the address? I have some gold plated USB and SPDIF optical cables to sell these people.

 

[C1]

Besides any re-mapping that may take place due to formatting, writing the data back onto the drive from another source effectively restores the data similar to if the HDD had been defragmented.

Not that I advocate the subject procedure, but doing so can help uncover drive and/or damaged file issues.

 

[Cerb]

That could probably have helped some in the early 90s, back when bad sectors were really bad, and you could do things like real low-level formatting.

A given location may get hard to read. The HDD will re-allocate this sector, and note how many it did in the SMART data. As a drive ages, this may occur multiple times. If it is a system drive, I like to go ahead and replace it if the count is above 0 (also consider that if someone is looking at the SMART data, they typically already have reason to believe something may not be quite right; or they want to be paranoid on the issue, since HDDs are cheaper than desk jockeys not having computers).

Any that weren't caught by the HDD, or were corrupted by other means, will be caught by the OS. Most common FSes today--any that would be used for business storage--perform CRC checking on each sector/block/cluster, when read. While you can't fix such an error in place, usually, that's what you have backups for.

A full format by Windows will check for bad blocks, as well, though these days, it will never find one. If it does, something else is very wrong, because HDDs come with plenty of re-allocation space (undisclosed, but I've never seen a healthy old HDD w/ even 10 actually re-allocated, so it need not be too many). It's a common sign of (a) unstable overclocking, (b) bad power, usually PSU or mobo, or insufficient chipset cooling (usually not an issue, today).

 

[VirtualLarry]

While I don't believe that the operation effectively increased the lifespan of the HDDs, I CAN see it extending the lifespan of the data stored on the HDDs, by re-writing the magnetic signals, that may have started to fade over the years. This is for really long-term stored data though.

 

[Phynaz]

How big are their "storage" drives? Honestly $83 per machine seems low for that much scut work.

Can you get the address? I have some gold plated USB and SPDIF optical cables to sell these people.

Automation, just do it with a boot USB. Probably 30 mins a machine tops.

 

[Gunbuster]

Not to mention possibly introducing bit errors by copying off over usb and then back. You would want to crc all the data, transfer, format, transfer, and then check the crc. But what about if you did detect an error. Really you need to generate checksums, transfer, check to make sure the transfer checksum matches, format, transfer back and verify the checksum again. Snore. Hours and hours per drive if you want to do it "right"

 

[Elixer]

All machines had 1TB drives, and they were all about 10-20% full.
He finished the job in a little under 6 hours, since he had to wait for a few people to "finish" their work.

 

[tweakboy]

That is gossip. The more you format the more possibilities of death for the HD. gl

 

[Gunbuster]

Each machine has a 1TB "storage" drive that is only 20% full? Who are these boneheads? Centralize the storage, get a backup strategy for that, and stop worrying about some bizarro FUD that drives will get bit rot...

 

[Coup27]

Completely unnecessary.

Edit: If each drive was 1TB and he finished in about 6 hours then he couldn't have possibly have done a full format on each drive which was the only reason for doing this in the first place. A quick format will just plonk a new file system down and will not scan for bad sectors or achieve any of the so called maintenance features he paid all this money for.

 

[VirtualLarry]

Completely unnecessary.

Edit: If each drive was 1TB and he finished in about 6 hours then he couldn't have possibly have done a full format on each drive which was the only reason for doing this in the first place. A quick format will just plonk a new file system down and will not scan for bad sectors or achieve any of the so called maintenance features he paid all this money for.

Uh-oh. Better call ripoffreport.com.

 

[Elixer]

No idea which Einstein setup what they have now, the database was the same on each system, by design.
There was no need to backup anything.
*edit, make that, 28 machines didn't need backup, 2 did, since they were not part of this insane design they got.

What they had is, when they are done for the day they run a program (script) to copy that data back to the main database from each machine, which then updates each machine's database for the next day at a set time of midnight.

Really, really bad way to handle data this way, in fact, this is perhaps the most idiotic system I have ever heard of.
As he was explaining all this, he was laughing so hard, at their ineptitude.

As for the how, he brought with him a few 12 bay aleratec units, and some other gear to copy/clone or erase.
He even told them their current setup wasn't very efficient at all, and again, there was no need to do this.
They just told him, it did not matter, this whole project was being funded from grant $$$.

Yeah, makes me sick as well.

 

[BrightCandle]

I don't doubt that bits not rewritten on a disk will slowly degrade. But I honestly expect the process to take dramatically longer than the drives useful life. A better strategy for dealing with this is to maintain backups and CRC check the files. That way you could happily check the contents of the disk in a fraction of the time, automatically and daily if you wanted without needing to do this nonsense.

If every bit matters then having a strategy to ensure you catch potential errors is what you need, not a rain dance.

 

[UaVaj]

Completely unnecessary.

Edit: If each drive was 1TB and he finished in about 6 hours then he couldn't have possibly have done a full format on each drive which was the only reason for doing this in the first place. A quick format will just plonk a new file system down and will not scan for bad sectors or achieve any of the so called maintenance features he paid all this money for.

x2.

30 machine with 1tb drive - all in 6 hours? it is possible if all the stars are aligned. (1) he super efficent, (2) has a dedicated backup drive for each machine, (3) no snag or technicial difficulties, (3) image backup the drive back to back, low-level format the drive back to back , image restore the drive back to back.

no mention of this - so chances are the client did not get what he paid for.

 

[UaVaj]

as for "why" one should even do this as interval maintaince.

low level format do physically check the drive for any degration.
image backup and image restore do refresh magnetic signals.

in this day and age. easier to mirror backup to another storage. should primary drive die. just buy a replacement drive and restore. no need to spend $2500 on preventive maintaince when a 1tb drive is ~$70.



one thing for sure this does not extend the life of the physical drive. in a sense it actually less the life from unnecessary task completed.

 

[Phynaz]

as for "why" one should even do this as interval maintaince.

low level format do physically check the drive for any degration.
image backup and image restore do refresh magnetic signals.

It's been a couple of decades since you could low level format a drive.

They are talking about a high level OS format, which is useless.

 

[Elixer]

x2.

30 machine with 1tb drive - all in 6 hours? it is possible if all the stars are aligned. (1) he super efficent, (2) has a dedicated backup drive for each machine, (3) no snag or technicial difficulties, (3) image backup the drive back to back, low-level format the drive back to back , image restore the drive back to back.

no mention of this - so chances are the client did not get what he paid for.

http://forums.anandtech.com/showpost.php?p=34677124&postcount=15

 

[UaVaj]

you got it confused.

although the practice is not done much anymore, there are still available utilities to low level format (try google) - should the situation deem necessary - such as this case in this thread.

as with a new drive purchase. there is a level of trust toward a manufacturer. hence no need for low level format. as with a used drive purchase. you definitely want to low level before putting your precious data on it. some of us will low level format a new drive just for our personal peace of mind.

low level format is basically zero-ing out the drive. while the platter is being zero out, it also gets checked in the process. a bad spot will not zero out and is marked bad.

once the drive get zero out. the next step is partition setup.

after that. then the high level format (aka quick format). which is the allocation table (FAT/FAT32/NTFS/etc).

now the drive is ready for data. all new data written will have a fresh magnetic signal.

 

[VirtualLarry]

you got it confused.

low level format is basically zero-ing out the drive. while the platter is being zero out, it also gets checked in the process. a bad spot will not zero out and is marked bad.

A true low-level format is not just writing zeros. Far, far more than that. Suffice to say that you cannot LLF a modern HD in the field these days.

 

[Cerb]

you got it confused.

although the practice is not done much anymore, there are still available utilities to low level format (try google) - should the situation deem necessary - such as this case in this thread.

Tools to brick your drive? I'll pass.

Low-level formatting was actually formatting the drive, arranging info about its layout in a way that the drive controller you happened to use liked (and/or OS version). Today, the disk controller in question is on the drive, it's formatted at the factory, likely in a format specific to that controller and firmware, and you don't get access. You have a black box view of a series of 512B and/or 4K blocks.

Since we went LBA, we have not had the ability to perform such actions, and it's been fantastic. The last time I did LLF, I couldn't afford a color VGA card .

 

[UaVaj]

was trying to keep it being simplfied.

initial low level format was the setup arrangement of the platter layout. done once set to maximum usable platter capacity. as it checked for maximum platter capacity, it also checked for magnetic quality of the platter.

future low level format can: retains the original platter layout or change the platter layout, net effect is zeroing out the platter. (which as what the job in question here requested to be done).