DIY HDD platter swap?

[edavid]

Please use server/RAID disks for servers and not desktop disks! Desktop disks are not designed to run 24x7 with a typical server usage profile.

Is there any evidence that server/RAID drives have lower failure rates?

Does over 15 years experience count?

No, I don't think anecdotal evidence has any value when it comes to disk drives.  It has to be actual statistics.

[sync]

No, I don't think anecdotal evidence has any value when it comes to disk drives.  It has to be actual statistics.

Sorry, I didn't done any statistics. I only have experience with hundreds of server drives and about 100 desktop drives in servers or RAID systems. And the differences are dramatic.

[codeboy2k]

Done that, successfully, just remember to also swap the SOIC-8 EEPROM and its all good(when its a pcb related damage, in the case, it was a grilled motor driver).

Well I think when I get around to it I'll attempt a head swap VERY CAREFULLY and when it's all closed up again I'll do a PCB swap too. So the bad drive will get new heads and its corresponding PCB. 

Without knowing exactly what is on the EEPROM (is it head data?  is it platter data? is it BOTH ? ) then I don't know if this will even be successful for me. But I'm willing to try.  I'm rebuilding my server with new drives first, and I'll attempt to repair and clone the bad drives later.

[codeboy2k]

The capacity of a LTO-3 tape is 400GB, compressed up to 800GB. You are taking about a few TB of storage. If you want to backup that you need a lot of tapes. Which means swapping tapes over hours for a full backup.

Yes I know, an LTO-3 drive is about $200 and tapes cassettes are cheap ($15) , so I can easily get several tape drives and do weekly and monthly backups without overwriting them.

I would prefer LTO-5 or 6, but to do the same with LTO-5, 6 is going to cost upwards of several thousand dollars.

Hmm.. I just took a look online at some retail stores, I can get a 4TB WD external USB 3 drive on sale right now for $149. At that price, it trumps everything.  I can get 2 for $300 and backup till the cows come home.  Gonna do it.  I think these are the ones with the embedded USB 3 interface on the drive electronics, it's not a SATA internal with a USB3 bridge card. So it's cheap.
 

[edavid]

Well I think when I get around to it I'll attempt a head swap VERY CAREFULLY and when it's all closed up again I'll do a PCB swap too. So the bad drive will get new heads and its corresponding PCB. 

Do the PCB swap first... much lower risk of killing the drive.

[edavid]

Hmm.. I just took a look online at some retail stores, I can get a 4TB WD external USB 3 drive on sale right now for $149. At that price, it trumps everything.  I can get 2 for $300 and backup till the cows come home.  Gonna do it.  I think these are the ones with the embedded USB 2 3 interface on the drive electronics, it's not a SATA internal with a USB3 bridge card. So it's cheap.

It's only 2.5" external drives that have integrated USB controllers.

[codeboy2k]

I also follow a "zero tolerance" policy on reallocated sectors - if any new ones develop after the drive has been in use, I replace the drive since more are coming.

Juding by your and other comments I think I should also have zero tolerance to reallocated sectors from now on. Seems like I might have saved my data if I already had followed that practice.  My tolerance level has always been higher, assuming that disk technology is difficult, and remapping of sectors is a way to get around these difficulties, but it doesn't mean my data is lost.  Even SSDs do this today, too.

From here out, I'll take a zero tolerance to any growing remapped sectors.

I also read online that some people put new drives through a grueling 2 or 3 day read/write stress test and check the SMART data before and after. If there is any re-allocated sectors or read/write failures during the tests, they RMA the drives.   Maybe I should do that too.

[codeboy2k]

Do the PCB swap first... much lower risk of killing the drive.

OK.  If that doesn't work then I'll try to bring over the head stack.

There's also some serial port commands that I'd really like to know what they are and what I can do.  I might just be able to bring the old drive back to life without swapping anything.

[codeboy2k]

Hmm.. I just took a look online at some retail stores, I can get a 4TB WD external USB 3 drive on sale right now for $149. At that price, it trumps everything.  I can get 2 for $300 and backup till the cows come home.  Gonna do it.  I think these are the ones with the embedded USB 2 3 interface on the drive electronics, it's not a SATA internal with a USB3 bridge card. So it's cheap.

It's only 2.5" external drives that have integrated USB controllers.

It is the WD Elements 4TB.  I think it might be 3.5". This is the one I mean:

From the USA (with free shipping to Canada)
$109 USD  + no TAX  http://www.bhphotovideo.com/c/product/1077555-REG/wd_wdbwlg0040hbk_nesn_4tb_elements_external_hd.html

In Canada (in my city, actually!) (also free shipping )
$149 CAD + TAX  http://www.ncix.com/detail/wd-elements-4tb-desktop-external-bb-103285-1145.htm

[madires]

Please use server/RAID disks for servers and not desktop disks! Desktop disks are not designed to run 24x7 with a typical server usage profile.

Is there any evidence that server/RAID drives have lower failure rates?

If you are looking for statistics, a quick web search will provide you with data. And I suggest to read some datasheets for desktop and server/RAID harddisks. It's not just about failure rates. The slow error recovery process of a desktop disk causes delays/timeouts in a RAID system, for example.

[madires]

No, I don't think anecdotal evidence has any value when it comes to disk drives.  It has to be actual statistics.

Sorry, I didn't done any statistics. I only have experience with hundreds of server drives and about 100 desktop drives in servers or RAID systems. And the differences are dramatic.

I fully concur. The differences are also quite visible in low performance, i.e. SOHO, applications. Desktop disks are designed to be cheap and to last a lot of start/stop cycles, not to run 24x7.

[wraper]

No, I don't think anecdotal evidence has any value when it comes to disk drives.  It has to be actual statistics.

Sorry, I didn't done any statistics. I only have experience with hundreds of server drives and about 100 desktop drives in servers or RAID systems. And the differences are dramatic.

I fully concur. The differences are also quite visible in low performance, i.e. SOHO, applications. Desktop disks are designed to be cheap and to last a lot of start/stop cycles, not to run 24x7.

A lot of Enterprise/Consumer models have identical hardware, the only difference seem to be in firmware. So with such models there will be a performance difference because the firmware is optimized for different tasks, but being more reliable is very doubtful. Seagate 7200.11 and ES.2 had exactly the same reliability (crap hardware) issues and even shared most of the firmware bugs (one of them caused HDD to lock if bad sectors appear).

[sync]

I fully concur. The differences are also quite visible in low performance, i.e. SOHO, applications. Desktop disks are designed to be cheap and to last a lot of start/stop cycles, not to run 24x7.

Yes. Seagate rates it's desktop disks for 2400h/year (~6.6h/day). So running them 24/7 is outside of their specification. It's like running an electronic component above it's ratings.

That's is one problem I have with the backblaze.com statistics. They should only shown disk which are rated for 24/7.

[madires]

A lot of Enterprise/Consumer models have identical hardware, the only difference seem to be in firmware. So with such models there will be a performance difference because the firmware is optimized for different tasks, but being more reliable is very doubtful. Seagate 7200.11 and ES.2 had exactly the same reliability (crap hardware) issues and even shared most of the firmware bugs (one of them caused HDD to lock if bad sectors appear).

I found following in the datasheets:
7200.11: MTBF 750,000h, high-performing, eco-friendly desktop storage
ES.2: MTBF 1,500,000h, business-critical Tier 2 enterprise drives, designed for 24x7 operation

[wraper]

A lot of Enterprise/Consumer models have identical hardware, the only difference seem to be in firmware. So with such models there will be a performance difference because the firmware is optimized for different tasks, but being more reliable is very doubtful. Seagate 7200.11 and ES.2 had exactly the same reliability (crap hardware) issues and even shared most of the firmware bugs (one of them caused HDD to lock if bad sectors appear).

I found following in the datasheets:
7200.11: MTBF 750,000h, high-performing, eco-friendly desktop storage
ES.2: MTBF 1,500,000h, business-critical Tier 2 enterprise drives, designed for 24x7 operation

And shows nothing more than marketing, of course Enterprise drives will have better MTBF (on paper).
Yet MTBF is quiet meaningless figure as it doesn't show life expectancy at all, just percentage of devices expected to die in a predefined time period.
Nonetheless both of those series were plagued like hell, and that MTBF was completely off unless they tested that MTBF for like less than a week.
BTW: http://knowledge.seagate.com/articles/en_US/FAQ/174791en?language=en_US

Seagate is no longer using the industry standard "Mean Time Between Failures" (MTBF) to quantify disk drive average failure rates. Seagate is changing to another standard: "Annualized Failure Rate" (AFR).

[edavid]

Please use server/RAID disks for servers and not desktop disks! Desktop disks are not designed to run 24x7 with a typical server usage profile.

Is there any evidence that server/RAID drives have lower failure rates?

If you are looking for statistics, a quick web search will provide you with data.

The only data I found indicates that desktop drives are more reliable: https://www.backblaze.com/blog/enterprise-drive-reliability/

And I suggest to read some datasheets for desktop and server/RAID harddisks. It's not just about failure rates. The slow error recovery process of a desktop disk causes delays/timeouts in a RAID system, for example.

That's more of a bug intentionally introduced into desktop drives.  Of course it's not relevant to non-RAID systems.

[sync]

The only data I found indicates that desktop drives are more reliable: https://www.backblaze.com/blog/enterprise-drive-reliability/

From the first link of the article (https://www.backblaze.com/blog/how-long-do-disk-drives-last/).



Interesting graphs. In my experience the failure rate of enterprise drives after 5 years is <10%. The 2nd graphs shows that Backblaze already had higher failure rates after 3 years with their consumer drives. And not all of them are desktop drives. They also using better ones and they selecting the better models.

Furthermore these graphs agree with my opinion to replace consumer drives after two years in 24/7 operation as precaution.

[Richard Crowley]

Just another way of drawing the famous "Bathtub Curve":

The 'bathtub curve' hazard function (blue, upper solid line) is a combination of a decreasing hazard of early failure (red dotted line) and an increasing hazard of wear-out failure (yellow dotted line), plus some constant hazard of random failure (green, lower solid line).

[edavid]

Interesting graphs. In my experience the failure rate of enterprise drives after 5 years is <10%. The 2nd graphs shows that Backblaze already had higher failure rates after 3 years with their consumer drives.

They addressed that:

It turns out that the consumer drive failure rate does go up after three years, but all three of the first three years are pretty good. We have no data on enterprise drives older than two years, so we don’t know if they will also have an increase in failure rate. It could be that the vaunted reliability of enterprise drives kicks in after two years, but because we haven’t seen any of that reliability in the first two years, I’m skeptical.

[helius]

The argumentum ad ignorantiam should be easy to spot.

[codeboy2k]

So to get my RAID server back up I decided to just go with 2 x  3TB disks in a simple mirror.

I bought the WD Red 3TB drive, WDC WD30EFRX-68EUZN0, and a Toshiba 3TB DT01ACA300

After 72 hours of continuous write/read/compare patterns of 0x55,0xAA,0xF0, 0x0F, 0x00 it was all fine on both drives.  Smart data showed 0 for the important metrics, and on both drives.  So I feel safe to use these drives.

Code: [Select]

196 Reallocated_Event_Count 0x0032   100   100   000    Old_age   Always       -       0
197 Current_Pending_Sector  0x0022   100   100   000    Old_age   Always       -       0
198 Offline_Uncorrectable   0x0008   100   100   000    Old_age   Offline      -       0

However, the WD drive had UDMA_CRC_Error_Count = 36 , while the Toshiba drive had this value at 0 still.

Should I suspect a cable problem?  I'm currently swapping the cable and re-running the tests to see if the CRC errors go up on the other drive now, but the tests takes 9 hours to complete 1 pattern. However,  with 36 CRC errors in 72 hours I expect to see 1 new error every 30 minutes

So it could be a cable, or a drive, or a controller problem.. If it moves with the cable I'll know. If it stays with the drive it could be a controller or a drive problem, so I'll switch controllers then to rule that out. 

[knks]

I remember reading somewhere - putting the HDD to fridge/freezer may make it readable 

[helius]

Likely an urban legend from the days when stiction was a common problem. A drive with stiction can't turn its spindle motor because the heads are cold welded onto the surface. If the bond forces between the head and platter are broken, the drive will at least spin up and may work. Freezing worked on the principle that the heads and platters would thermally contract along different directions. Another technique was tapping the drive with a hammer while switching on power.

Stiction stopped being an issue when head assemblies were redesigned to prevent it. Freezing is unlikely to help now, and could cause condensation which quickly corrodes the data surface.

[codeboy2k]

Freezing doesn't work with new drives.. I'm not even going to try it.

I'll do a head swap when I get time to make a portable "clean-box" .

As for that last post of mine, where I said UDMA_CRC_Error_Count = 36 on one of the new drives,  I swapped cables, swapped controller ports, but nothing new happened and nothing changed after days of additional read/write testing. That number stayed at UDMA_CRC_Error_Count = 36 and the other drive never gained any errors when it was swapped around too.

So I think that was an anomalous event that caused the controller attached to that specific drive to briefly screw up, but I'm not seeing it again.

Today, both drives are in service now, and it's still UDMA_CRC_Error_Count = 36 for that drive and 0 for the other drive.

[CatalinaWOW]

Joining this party very late, but had to comment.  Codeboy you say that this is just personal data "not worth recovering".  I would encourage you to think long and hard about this before you do anything that reduces your chance of recovering the data.  The type of data you describe represents many, many hours of your involvement.  Capturing, digitizing, interviewing, whatever.  When you say it isn't worth recovering you are placing zero value on that time, and also zero value on any future time you may spend attempting to re-create portions of the lost data.

I am speaking from personal experience, having lost a huge chunk of personal data to a moment of stupidity.  I went ahead and spent some serious dollars on recovery, and while there were still many things lost, I have never regretted the money spent.  I have rued the inadequate backup strategy, the inadequate planning that resulted in the data loss and other things, but not the investment in recovering the data.  In my case the investment (after converting dollars to hours) was less than ten percent of the hours originally spent getting the data onto the hard drives.