Backup bods Backblaze: Disk drive reliability improving Cloud backup bods Backblaze this week cracked open another spreadsheet and fed it some log files, spitting out a bunch of fresh drive availability stats that show a mild uptick in Seagate drive failures. Over the last three years, the general trend for both Seagate and HGST has been that annualised failure rates (AFR) have …
I think that statement is misleading given the small number of drives and drive/days it has - which indicates these drives have been in service LESS THAN A MONTH during this 90 day snapshot. The 12 TB Seagates with the 2.22% failure rate are clearly worse.
Drive failures follow bathtub curve, with more failures shortly after going into service, then a long period of relatively low (for the model at least) and stable failure rates, and eventually an increasing rate (which is probably never an issue for Backblaze as they aren't likely to keep drives in service long enough to reach this point)
Do they replace the drives at the first S.M.A.R.T. pre-failure warning, or when they actually crash? What if different makes of drive sense pre-failure differently? Sometimes all a drive has to do is overheat slightly from intense activity. Custom designed drive racks could influence this, and perhaps Helium drives dissipate less heat?
If you were a drive manufacturer, would you set the pre-failure criteria to be more sensitive, so customers buy more replacement drives?
I've never been treated more ludicrously or poorly than I was by the techs at Backblaze. They lost track of my encryption key and attempted to blame it on the user, that being me. I still have my key! Thanks for losing my data then telling me to get lost and never return. It's no surprise that I consistently run into people on the net who've suffered from the same experience.
Look elsewhere for cloud storage, please.
Many moons ago I started to take HDD cooling more serious, keeping them below 45°C instead of > 55°C. The failure rate dropped down to "exchange the drive because it is too small" instead of "cause it failed". Today, even on a warm day, my drives don't go beyond 36°C. Including some ST12000NE0007. And the smaller, much slower though, 3 TB Red drives are at 30°C.
Cooling HDDs makes a big difference.
Only those Seagate shingled magnetic 8 TB Archive HDD's were no good, they failed way too often no matter what you do.
Ah yes it does, back many years ago I was proud to have upgraded my gaming PC with 2 Maxtor DiamondMax (2 I think) 320GB 7200 rpm drives.
Then after having some performance issues, I did a Virus Scan thinking something dodgey was going on. Several hours later the machine had fallen over and the BIOS failed to detect either drive. I will never forget opening the case and realising how hot the drives had got. After trying to figure out what happened, I noticed realised to late that my case had no airflow over where the drive cage was mounted, I then made this worse by putting them within a cm of each other when I could have spaced them out.
"It shows a general trend of improving AFR numbers with several points clustered along a high left to low right diagonal line."
That's an OK general conclusion, but if you look brand-by-brand, it's not quite so simple:
1. First, WD has only one drive model in the study, so we can't really tell how well they're doing.
2. Second, Toshiba has gone from 0% AFR (4TB) to 0% AFR (5TB) to 0.33% AFR (14TB). This is mitigated by the fact that that 0.33% is caused by 1 drive failure -- the only failure at all for Toshiba in the entire study. So it's possibly an artifact of the sampling rather than an actual decrease in quality.
3. Third, HGST does indeed have a 2.6% AFR for their LE600 models, but that's due to 1 drive failure as well -- that might, like with Toshiba, be an artifact of the sample. Setting that aside, HGST has 0.23% and 0.34% AFR for their two 4TB models, 1.17% for their 8TB model, and 0.56% for their LN604 12 TB model. This is not a high-left to low-right diagonal line, but a hill, indicating a decrease in quality for HGST around the time they were producing their 8TB models which was (mostly but not fully) resolved with the LN604 model.
4. Finally, Seagate starts high in the high-left (1.96% AFR for their 4TB model), improves for their 8TB model (0.27% and 1.64%) and 10TB model (1.01% AFR), and then totally implodes with their 12TB model (2.22% AFR). They are the closest to a high-left to low-right line, but they lose it at the end.
So the only reason the high-left to low-right line exists is because models from different brands (Seagate, WD, HGST, and Toshiba) are clustering near that line. But since the brands have their own production facilities, this is most likely a coincidence and not representative of a meaningful trend.
Looking brand-by-brand, we see:
WD underrepresented;
Toshiba holding near-perfect but possibly having a bit of trouble with their 14TB drives;
HGST starting strong but having problems starting with their 8TB models possibly through to the 12TB LE600 model, then improving again with their LN604 model; and
Seagate steadily improving quality from their 4TB models through to their 10TB model, then getting worse with their 12TB model -- all the while maintaining the worst overall record (yes, the HGST LE600 model has a higher AFR, but that's due to a single drive failure as mentioned above, so is likely inflated.)
As for the 12 TB Seagates: It only applies to that specific model. Other Seagate 12 TB drives may be much more reliable than those enterprise drives.
I remember, when 500 GB was common, the WD 500GB RE2 Enterprise drives died like magflies. Real death, click-click-click style. While all other WD 500GB drivers were solid. In all those servers which those specific drives we soon asked specifically for Seagate 500 GB replacements instead of WD 500 GB RE2 since the failure rate was near "one each week" for less than 50 drives.
I've been reading the BB reports for years, and Seagate have been by far the worse make for all but one of those years - the year WD had all those issues.
EVERY Seagate drive I have ever owned - failed young, yet nearly all of the other brands - including WD lasted until the point you could buy cheap SD cards with a higher capacity.
I have had 1 HDD drive fail in the last 10 years, an 8 y/o 3TB Toshiba; leaving me with 3 of the same model still going.
In the last 3 years I have had FOUR SSD drives fail, from 3 different brands.
In friends computers, most were supplied with Seagate drives, and most needed replacing with new dives due to failure within 2 1/2 years; Seagate wouldnt honour warranties.
One Hitachi drive failed after someone kicked the crap out of their PC, Hitachi replaced it without a quibble.
That's only part of the story though. HGST was acquired by WD however, part of the acquisition was that it had to operate independently. I think that has now changed and HGST are properly consumed by WD and that the HGST brand will disappear (however, with the time line in the BB article, I believe the HGST drives were HGST (although owned by WD)).
"That's only part of the story though. HGST was acquired by WD however, part of the acquisition was that it had to operate independently. I think that has now changed and HGST are properly consumed by WD and that the HGST brand will disappear (however, with the time line in the BB article, I believe the HGST drives were HGST (although owned by WD))"
Other views are available but I really wish that HGST had not been bought by WD because now I don't know how long the higher standards will last under the current owner.
"Not sure if it is still true, but all of the Toshiba drives I have bought internally refer to themselves HGST drives."
That's because they _ARE_ HGST drives. Lookup the merger and divestiture trails.
Toshiba acquired the facility as part of agreements imposed by Chinese regulators to prevent a duopoly forming with consequent market abuse issues (which we'd already seen post-2011 floods)
"part of the acquisition was that it had to operate independently."
That was a condition of merger imposed by the Chinese anti-monopoly regulators. Seagate faced similar restrictions and there were requirements to divest parts (which is how Toshiba strangley ended up as a major 3.5" HDD player years after effectively exiting the market in favour of 2.5" format devices.)
My summary when the Chinese regulators finally allowed HGST to be folded into WD was that SSDs were at the point of parity with HDDs and a HDD duopoly no longer posed a threat to the market. Subsequent events have proven me correct - and the rapidly falling price of multi-TB SSDs is underscoring that summary.
It doesn't matter if they 'only' have 1000 rewrite capacity, when a 16TB "archival" drive is write once, read occasionally - and I'm waiting to pull the trigger on my aging spinners as it'll save a several kW of power consumption along with precious rack space.
My experience for years has been that the preferred brand list has been:
1) HGST
2) WD (Now owns above)
3) *
4) Seagate
I’ve actually had better luck with old IBM “DeathStor” drives than a lot of Seagates.
As for consumer vs enterprise, the low-end consumer are crap. The high-end consumer and enterprise are pretty much the same, except for the whole SATA vs SAS mess. Usually even the SATA versions of high-end consumer and enterprise are roughly the same.
1) HGST - agreed
WD and Seagate take turns producing utter crap. Seagate's STx000DMx ranges being something to keep far FAR away from with multiple failures within the warranty period being common (the earlier DLs were slightly better, subsequent models have so far proven much better)
_BOTH_ HD and WD have grown by mopping up other HDD makers which went bust after producing bad runs of drives - and in any case they're now all using the same platters + heads, so the secret sauce is in the motor bearings and software.
Someone please explain to me why an extra screw in the top of the case (supporting the spindle at both ends) is worth an extra $50-150 when the non-supported spindles have threaded holes for the screw when you lift the cover off.
Oh yes: "Because they can" - and that shows how cynical the manufacturing of HDDs has become.
Well, the consensus has been that HGST manufactured better HDDs. WDs were usually OK, but go with the higher performance Blue and Black models and stay away from the Green and Red models. Toshiba was also OK, although they didn't have the reputation of HGST or the popularity of WD. Seagate, apart from certain models, always seemed like a crap shoot. Personally, I mostly run HGST refurbs in my storage servers. They may be a little noisy but they are in the basement where it is cool year round and I don't care about the noise.
The thing about HDDs is nobody will pay what it would cost to buy an HDD guaranteed to last 10 years. That said, manufacturers build HDDs to run close to edge of failure in order to cram as much storage (up to 16TB) as they get into a 3.5-inch form factor using tricks like Helium and SMR. Areal density of 1.x terabits per square inch is the limitation with current HDDs. When HAMR and MAMR HDDs appear they will have even higher terabit per square inch values and probably be more expensive to produce, which means they will cost more per GB. Meanwhile, SSDs are getting larger and cheaper. SSDs rule the 3TB and under storage market. HDDs will make their stand in the capacity storage market until they are eventually challenged by SSDs.
Owing to the lack of availability of HGST drives these days I ended up going with the Seagate Ironwolf. I had moved to a WD Gold but found it ran way too hot - 5-6 degrees above the surrounding HGSTs. Red Pro were quite expensive and I didn't want 5400RPM (mixed spindle speeds) or Intellipower (had circuitry fail in the past) drives.
Ironwolf going ok and figured after at least 7 years of steering clear it was worth another shot. Time will tell. Toshiba looking like an option but availability is not as great. Seagate are really pushing the new branded drives.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
