Hard drive reliability

[Pentium4]

I have a shit ton of used hard drives, most of them working. I love checking the POH, age at time of failure, etc.

I have noticed that old drives are MILES better than the garbage being pumped out now a days. The 2008-2010 Western Digital Blue drives and Seagate 7200.9/10 drives are the most reliable as far as non server stuff goes.

I've also noticed IDE drives are pretty reliable, seeing many pass the 50,000 POH mark. I have a Seagate Barracuda ATA IV that has about 96,000 POH with no bad sectors! And 3 IBM drives with over 83,000 hours.

Is it true that they were just made better back then?

2013 1TB 2.5" Toshiba drive made it over 20,000. Being used as a backup drive for my brother now.




This 320GB 2.5" WD drive did pretty good. It was in a laptop and had over 4,000 power cycles. Most of the drive was good. It only had 7 bad sectors and my HDD Regenerator program could not fix them. It was bad enough that it would freeze the OS that was on the drive. Overall I'd say it did its job, considering most people are stupid with laptops and have no sense of treating things with care.




I've never had good experiences with Toshiba drives but this one showed me they're capable of making a decent drive. This was in a laptop made in early 2010 and it surely had a useful life span. It had bad sectors at 7% into a drive scan, but upon trying to repair them, I heard it fail completely! I heard some really loud clicks and the drive spun down never to spin up again. I gave it a pat on the back and told it that it did a good job





This thing just sucks....Didn't even make it to 8,000 hours. Was even in a laptop being used as a workstation so it was almost never moved. When trying to repair the disk, it would freeze the computer. Awful...




This thing was in a server deployed in early 2003. It ran about 24/7 into mid/late 2012. It's a 182GB IBM IDE drive. I plugged it into a Socket 423 system that I built in my office just to host files/tools that I use. I took that picture on 3/30/17 and actually hasn't been power cycled since, so it's almost at 87,000 POH now!




This is the drive I have in my server just to store my CDs ripped into uncompressed WAV files. It's a 250GB Western Digital Black 2.5" drive. It was in a 1U server, so that's how it's lasted so long. I rarely have faith in 2.5" drives but if you treat them well, I guess that helps. This one was made on June 12, 2011.

[Pentium4]

Re: Hard drive reliability

Here's a Seagate drive that's lasted a while.

[Wester547]

Re: Hard drive reliability

Well, the Seagate Barracuda ATA IVs weren't perfect. I had one (a ST340016A) whose death was signaled by way of a horrible clack-of-death upon spinning up, after 16,000 POH / 7,000 power cycles (these are really quiet drives but it really made a racket when it failed). I'm guessing it has a bad HSA. It had 50 reallocated sectors, though. It failed four years ago, and was never subject to any operating or non-operating shock AFAIK. It was made in November 2001, in Singapore.

I have a WD5000BEVT, manufactured on December 8th of 2009 in Thailand, whose S.M.A.R.T. record is absolutely clean. It came from a Dell Inspiron 1764. It passed a full HDD scan, too, which means there are no bad sectors on it. It has 18,500 hours on it and 6,100 power cycles. So I'd say it did pretty good too. But of course it may be the exception to the rule, as an internet search reveals the WD5000BEVT to be a very failure prone drive. Part of that may be excessive non-operating / operating shock, but I get the sinking impression that the majority of 2.5" HDDs, especially those modern, are crap. Toshiba is no exception to that. How interesting that your drive surpassed 20,000 hours - that's the very spec they are rated for in accordance to the datasheet.

In answer of your question (if I'm wrong on any of the following, don't hesitate to correct me) - the data density on older hard drives was much lower. On a 40GB HDD with 2 heads and 1 platter, for an example, the data density is so low that it's possible for the HDD to fix and correct errors before you would ever know there was an issue. Compare that to a 1TB HDD with 2 platters and 4 heads, and not only does the data density increase dramatically (raising the chance of data corruption), but the target zones become smaller which means there is less margin for error - IE, the heads must write and read data to the platters in a far more precise manner (I believe this is part of the problem with perpendicular recording technology too, compared to the older longitudinal recording technology). Flying height is another issue. That's the distance between the heads and the platters, and this has become smaller and smaller as HDDs have become larger and faster, which makes a head crash a far greater likelihood. It also puts more wear on the HSA as thermal asperities will also arise as the result of the lower flying height.

There are other issues which decrease the reliability of modern drives but yes, I generally agree that older HDDs are leaps and bounds above the newer HDDs as far as reliability goes. That can be said of many other topics regarding modern computers, too. Modern GPUs and motherboards have a tendency to drop like flies for a myriad of reasons that belong in another thread.

[stj]

Re: Hard drive reliability

i have been testing drives this week for someone,
the newer ones are shit - no doubt about that.
if buying new, WD black, i dont know if they are good - but:
if they give a 5year warranty then they obviously put more attention into it's construction!

if salvaging used drives then SCSI are best - i have over a dozen that are very old and still good.
then IDE, then SATA last.
also,
stay away from used 3.5" toshiba drives.
i have about 10 - all dead.
and the diagnostic software - i use that term lightly, is fucking useless.
it just says "FAIL" in ASC-ART.
no info on why or any other usefull hints.

[Pentium4]

Re: Hard drive reliability

Some more reliable drives.

[Pentium4]

Re: Hard drive reliability

In answer of your question (if I'm wrong on any of the following, don't hesitate to correct me) - the data density on older hard drives was much lower. On a 40GB HDD with 2 heads and 1 platter, for an example, the data density is so low that it's possible for the HDD to fix and correct errors before you would ever know there was an issue. Compare that to a 1TB HDD with 2 platters and 4 heads, and not only does the data density increase dramatically (raising the chance of data corruption), but the target zones become smaller which means there is less margin for error - IE, the heads must write and read data to the platters in a far more precise manner (I believe this is part of the problem with perpendicular recording technology too, compared to the older longitudinal recording technology). Flying height is another issue. That's the distance between the heads and the platters, and this has become smaller and smaller as HDDs have become larger and faster, which makes a head crash a far greater likelihood. It also puts more wear on the HSA as thermal asperities will also arise as the result of the lower flying height.

I'd be curious to know how accurate this is!

i have been testing drives this week for someone,
the newer ones are shit - no doubt about that.
if buying new, WD black, i dont know if they are good - but:
if they give a 5year warranty then they obviously put more attention into it's construction!

if salvaging used drives then SCSI are best - i have over a dozen that are very old and still good.
then IDE, then SATA last.
also,
stay away from used 3.5" toshiba drives.
i have about 10 - all dead.
and the diagnostic software - i use that term lightly, is fucking useless.
it just says "FAIL" in ASC-ART.
no info on why or any other usefull hints.

What were the POH on the SCSI drives you've seen?
And yeah warranty says a lot about a company's faith in their product lol
They've all seemed to get lower as time goes on. I noticed old SATA drives are more reliable than new ones. I've had at least 6 2-5TB Seagate/WD drives fail at under 10,000 hours and they fail so bad that the computer gets stuck at the BIOS splash screen. On the other hand, I have three 2TB Seagate drives made in 2013/2014 and they're all about to hit 30,000 hours. I guess I got lucky.

[momaka]

Re: Hard drive reliability

In answer of your question (if I'm wrong on any of the following, don't hesitate to correct me) - the data density on older hard drives was much lower. On a 40GB HDD with 2 heads and 1 platter, for an example, the data density is so low that it's possible for the HDD to fix and correct errors before you would ever know there was an issue. Compare that to a 1TB HDD with 2 platters and 4 heads, and not only does the data density increase dramatically (raising the chance of data corruption), but the target zones become smaller which means there is less margin for error - IE, the heads must write and read data to the platters in a far more precise manner (I believe this is part of the problem with perpendicular recording technology too, compared to the older longitudinal recording technology). Flying height is another issue. That's the distance between the heads and the platters, and this has become smaller and smaller as HDDs have become larger and faster, which makes a head crash a far greater likelihood. It also puts more wear on the HSA as thermal asperities will also arise as the result of the lower flying height.

Yup, that about sums it up about reliability.

The only thing I would like to add is that a lot of RoHS-compliant HDDs, regardless of age, fail due to bad/corroded contacts between PCB and HDD head amps assembly. I saved the data for a few people simply by cleaning those. When these contacts go bad, they can make the HDD rack up bad sectors or get many Hardware ECC errors. But once the contacts are cleaned, the working condition of the HDD restored, and the drive zero-filled, it can look like a new drive.

I'd be curious to know how accurate this is!

Well, I am no expert either, but I think what Wester wrote above is just common sense about simple physics. So I'd like to think it is accurate. If you disagree, have a look at the durability between Blu-ray discs vs. DVDs vs. CDs. An average scratch on a CD may or may not cause an issue, but could easily cause a CRC on a DVD and many CRCs on a Blu-ray. And again, it all has to do with the data density - there is simply much more data on a Blu-ray disc, so even a tiny scratch could wipe some data. Of course, there is more to that than just the data density: DVDs use much more complicated CRC schemes than CDs and Blu-ray discs even more complex ones - they have to! Otherwise, we wouldn't even have that technology. IIRC, a DVD uses much higher percentage of its physical storage space for CRC code than a CD.

stay away from used 3.5" toshiba drives.
i have about 10 - all dead.
and the diagnostic software - i use that term lightly, is fucking useless.
it just says "FAIL" in ASC-ART.
no info on why or any other usefull hints.

lol
Sounds like a very professional software they got.

[Wester547]

Re: Hard drive reliability

Yup, that about sums it up about reliability.

The only thing I would like to add is that a lot of RoHS-compliant HDDs, regardless of age, fail due to bad/corroded contacts between PCB and HDD head amps assembly. I saved the data for a few people simply by cleaning those. When these contacts go bad, they can make the HDD rack up bad sectors or get many Hardware ECC errors. But once the contacts are cleaned, the working condition of the HDD restored, and the drive zero-filled, it can look like a new drive.

Ah yes, I forgot to mention the issue with RoHS drives experiencing corrosion and oxidization on their VCM and preamplifier PCB contacts. The contacts can actually still look good, and you could still experience the issue.

I worded some of what I wrote poorly since it was late when I typed that all up. I basically meant to say that modern HDDs have tighter tolerances because the bits of data written to the platters and read from the platters are significantly smaller than they were on older drives, which I think is the crux of what momaka was saying.

Well, I am no expert either, but I think what Wester wrote above is just common sense about simple physics. So I'd like to think it is accurate. If you disagree, have a look at the durability between Blu-ray discs vs. DVDs vs. CDs. An average scratch on a CD may or may not cause an issue, but could easily cause a CRC on a DVD and many CRCs on a Blu-ray. And again, it all has to do with the data density - there is simply much more data on a Blu-ray disc, so even a tiny scratch could wipe some data. Of course, there is more to that than just the data density: DVDs use much more complicated CRC schemes than CDs and Blu-ray discs even more complex ones - they have to! Otherwise, we wouldn't even have that technology. IIRC, a DVD uses much higher percentage of its physical storage space for CRC code than a CD.

IIRC Blu-ray Discs have an extra layer of polymer used to coat the surface, to protect them from scratching. It doesn't have a 100% success rate but it would explain why some Blu-ray discs still work despite some minor surface damage.

I think the upshot is that while not all modern drives are unreliable, they definitely have a higher failure rate than older drives, for all the above mentioned reasons.

[stj]

Re: Hard drive reliability

it's not because of RoHS,
it's because the tight bastards switched to silver-plate instead of gold.
it tarnishes.

[diif]

Re: Hard drive reliability

Density and cost. Far more data packed into a platter with most drives having multiple platters.
Manufacturers have always tried to reduce costs such as contact plating and testing (Seagate firmware bug ?) Probably other things internally too. It was kind of obvious where things were going when Seagate warranties dropped from 3 to 1 year. Maybe they lessened the quality to improve sales. "Hey we do tell you, so buy two so you can back up"
Hard drives used to come in expensive devices and were treated with respect. Anybody that fixes laptops will tell you that's not the case any more.
Luckily now we have SSDs and hard drives are cheap and big so we can all back up our data and not worry about loss.

[momaka]

Re: Hard drive reliability

IIRC Blu-ray Discs have an extra layer of polymer used to coat the surface, to protect them from scratching. It doesn't have a 100% success rate but it would explain why some Blu-ray discs still work despite some minor surface damage.

Yes they do.
Also, I think starting with DVDs, they also added a thin protective layer on the top of the disc as well, since the reflective layer where the data is stored is actually at the top side of the disc. That's why it's actually easier to damage a CD-R if you put a deep scratch on top of it than on the bottom, as CDs do not have that protective layer on top that DVDs and BRs do. It's also why they say never to use a writing utensil with a hard tip - it can damage the disc very easily. I've actually done that before once, when I tried label a disc with a pen. Needless to say, I ended up making a see-through scratch on the disc and some of the data on it got corrupted. But it was a CD-R disc, so didn't loose much data, lol.

Luckily now we have SSDs and hard drives are cheap and big so we can all back up our data and not worry about loss.

Yes, but new HDDs tend to be so unreliable that it's best to keep at least one back up of your backup, lol.

Also, regarding SSDs: back when I worked in the computer store/repair shop last spring, we were told to tell people to move to SSDs as they are more reliable due to not having any moving parts. But after several months, I quickly learned that is not exactly true. What's worse, when SSDs fail, the chances of recovering data are almost null. With a mechanical clunker, at least you can swap parts to get your data back. Heck, most anyone here who is a skilled tech and has enough patience can probably do a head swap himself/herself. It's not rocket science, and you don't absolutely need a clean room to do it. It's just that when you open up an HDD without a clean room, the chances are it will get more errors due to dust and is likely to crash sooner than later. But for backup purposes, it may work just enough.

[diif]

Re: Hard drive reliability

SSDs are oblivious to bumps and bangs but unlike hard drives that tend to let you know they are poorly due to their noise, SSDs just die. I got bitten by the Intel bug on one of their drives. Luckily it was a new build and in testing. There are plenty of cheap nasty SSDs and sticking with Samsung and Intel is a fairly safe bet.
Back up and then no need to worry about recovery. Online and/or offline.

[jarvis7]

Re: Hard drive reliability

lol.. i opened this thread and while reading i remembered i too have a Toshiba drive in my laptop which i had checked 10 days ago with crystaldiskinfo because my laptop warranty was expiring and again 4-5 days ago rechecked it on the last day of warranty

didn't look at power on hours then , so curiously i wondered what's mine like ..

i opened it up to check mine.....

andddd..., i was presented with Caution sign

1 Uncorrectable sector count and 8 current pending sector count .. just 5 days after warranty expired

12,700 - power on hours
3,916 - power on count
500GB Toshiba

laptop is a 2014 model , and iirc HDD was also 2014 made ( will open it up later )

my 2012 Samsung 2.5inch 500gb gave first caution sign at 9000 hours or something, later replaced in warranty

[eccerr0r]

Re: Hard drive reliability

I actually have nothing bad to say about Toshiba though I haven't had many of their drives. I have to knock on wood though, my current RAID array are 2TB Toshibas.

All in all, yes I had fairly decent luck with the early-mid 2000s 120GB disks. I had a mixture of Seagate and Maxtor disks, and other than some power connector issues, not a single disk died. A few of the seagate-maxtor disks exceeded ~72000POH and still would be kicking if I had not needed more space. Most got reassigned to be random OS/boot disks.

The 500GB disks that replaced/upgraded the 120G array were a disaster IMHO. I had a RAID of WD and HGSTs and the disks kept dropping like flies. I think the 4x500G disk array went through:
2 WD 500G SATA's (random)
5 HGST 500G SATA's
1 Seagate 750G PATA.

First disk that died was a HGST. RMA'ed right away (infant failure?)
Then another HGST died. Not sure if it was heat or what, but it died.
Then I had the two WDs die in sequence over a few years, which was really odd.
Despite the SATA array I had to run one PATA disk to keep it going as by then 500G disks weren't cost effective so I had to throw in a non identical spare, thank you Linux MDRAID.
The next disk that failed was another HGST, and the final straw that killed the array was a spurious failure of yet another HGST.
The HGSTs were the "main" disks and I was tossing in WDs as fill and I was running one hot spare for a while when I had 5 serviceable disks. A HGST passed the 67000POH mark before the array was toast.

I currently have all the disks I haven't tossed yet from that array with two 2T disks and one of the old 120G disks (boot disk) in the P4 machine with the bodged 750W PSU. This 750W PSU spun up all nine disks just fine. Now to see which disks are still serviceable and which belong in the trash...

Additional comment: It seems a lot of my disks now have rapidly decreasing SMART field #9 as they reach 60000 POH. I have one down to 13 at 64000 POH. The 120G disks with 70000 POH is only down to 31.

[stj]

Re: Hard drive reliability

HGST's were the Legendary IBM Deathstars - IBM sold the low-end production line to Hitachi.
(they kept the SCSI / SAS line, they werent stupid!)

[eccerr0r]

Re: Hard drive reliability

The first two HGSTs that failed for me were a desktop line and a server line. Granted that server line disk was a refurb so it already had many hours on it.
All my other disks are "desktop" disks - though the seagate was the first gen perpendicular record (7200.10 IIRC).

...

Dang, that WD disk that died on me is alive once more. I could not get it to detect at the time on several machines, not sure what happened.

[kevin!]

Re: Hard drive reliability

Some more reliable drives.

It retains very well those units , for me they are the best, they will not be manufactured units like these again.
By the way, I also love to check the POH in the units hahaha.
If you do not mind, I can share my units so you can see how they are .

I agree, the Seagate 7200.7, 7200.9 and 7200.10 series are the most reliable.
I have several of the 7200.10 and none shows problems, and have passed the 30,000h of the average, something that surprised me a lot, they are all IDE.
And I have a Maxtor IDE from Seagate's 7200.10 series of 160GB, it's 55,000H if I remember correctly (I've had it disconnected for a long time)
And laptop disk, the longest I've ever seen, was a hitachi of 2.5 of 160GB with 46,000h and had no flaw .
That yes, the HP notebook broke the plastic to clusters, although it worked very well for the hours it had, including the screen had a lot of brightness.

I withdraw this Seagate in May, just when I reached 100,000H began to relocate sectors, poor: '(.



Seagate 7200.10:

By the way, I'm glad to read you through the forum again Pentium4 .
Bye

[stj]

Re: Hard drive reliability

i just found a 1.2gig quantum fireball 1280AT in my pile,
made in 1996, and operating 12hours a day, 7days a week until about 4 years ago.
AND IT STILL WORKS - I JUST CLONED IT!!!!!

incase your wondering, it was in an Atari AREA51 arcade machine.
(maybe it was built using back-engineered alien technology! )

[kevin!]

Re: Hard drive reliability

i just found a 1.2gig quantum fireball 1280AT in my pile,
made in 1996, and operating 12hours a day, 7days a week until about 4 years ago.
AND IT STILL WORKS - I JUST CLONED IT!!!!!

incase your wondering, it was in an Atari AREA51 arcade machine.
(maybe it was built using back-engineered alien technology! )

It is very mythic those units of Quantum, I would like to return to hear those clicks that made when start hehe.
I had one of 3GB, and one of 1.2GB that died (though I still keep it).

[RukyCon]

Re: Hard drive reliability

The drive in my main PC [Which has been going for 3 years now] has 20K POH with a little less than 2000 POC. So far, no problems.

The drive is a Seagate HN-M101MBB. [Commonly sold under the Samsung name]