BackBlaze - HDD reliability stats for Q2 2015

[burnttoastnice]

Samsung left the HDD market in 2011.

 

Oh balls, I better be looking at replacing the Spinpoint HDD in my laptop with a SSD soon

[Bsmith]

When their first report came out I was in full support of it,  I think I even posted here about it and pointed out why people should accept when the test consists of such a large volume of samples. By the time the second came out I started reading all the blogs and reports from data companies and specialists in the field.  Now the third one has surfaced, literally nothing has changed in BB, more specialists are coming out of the woodwork to criticize their method, less information is available on their setup,  they still have a significant disparity between the number of branded drives and no control grouping.  I subscribe to the notion that if the fish has an odor it's probably off.

 

EDIT: ignore the control grouping, it's not really important for this type of study.

 

you shouldnt be serious on forums when just awake or you end up oing stupid stuff, not for myself.

well with you mentoining this I looked up the Q1 blog results, it's very interesting that the imagie in OP has diffrent failure rates at Q1 2015 compared to the original post in the blackblaze website.

for example, the first 2 drives HGST 7k2000 and HGST 5k3000 show 1.15% failure rate and 0.74% failure rate, while in the original article the rates are 1,66% for the 7k2000 and 0.79% for the 5k3000, for Q1 2015, so I'm not sure what they did or how they did recalculate the valvue's of this, but it seems weird that the numbers changed.

 

link to 2015 Q1 results:

https://www.backblaze.com/blog/best-hard-drive/

 

so far I could find only the WD 4 TB (WDC WD40EFRX) has the same failure rate in Q1 2015

Edited July 29, 2015 by Bsmith

[mr moose]

well with you mentoining this I looked up the Q1 blog results, it's very interesting that the imagie in OP has diffrent failure rates at Q1 2015 compared to the original post in the blackblaze website.

for example, the first 2 drives HGST 7k2000 and HGST 5k3000 show 1.15% failure rate and 0.74% failure rate, while in the original article the rates are 1,66% for the 7k2000 and 0.79% for the 5k3000, for Q1 2015, so I'm not sure what they did or how they did recalculate the valvue's of this, but it seems weird that the numbers changed.

 

link to 2015 Q1 results:

https://www.backblaze.com/blog/best-hard-drive/

 

so far I could find only the WD 4 TB (WDC WD40EFRX) has the same failure rate in Q1 2015

 

You were looking at their annual failure rate figures not the figures for the quarter.

 

The first graph is the total average for the year, while the second graph is the quarterly results.

[Bsmith]

You were looking at their annual failure rate figures not the figures for the quarter.

 

The first graph is the total average for the year, while the second graph is the quarterly results.

 

whoops! guess it's still to early for me.

[mr moose]

whoops! guess it's still to early for me.

 

It happens to the best of us.  

[Prophet of Entropy]

i wonder why wd's numbers are worse than hgst's eventhough it's an "wd company."

something to do with HGST drives costing 2-3 times as much as the WD drives.

[Psykomantis00]

i wonder why wd's numbers are worse than hgst's eventhough it's an "wd company."

HGST was bought by WD. They aren't the same thing. It used to be owned by Hitachi. 

[ChineseChef]

-snip-

 

 

-snip-

 

 

-snip-

 

 

The problem is not the method or the data.  The problem is people think this is some kind of study or test.  There is no testing, no studying here.  It is just essentially random numbers released by some company.  They simply state that this is what they use.  Then everyone and their mother thinks it is some kind of professional study done to determine the efficacy of HDD makers.  I will say that the way they put out the data is a bit misleading, but that is mostly for the very purpose that it causes controversy, thus spreading their name around, its basically just free publicity. 

 

The data is not flawed in and of itself.  It is 100% accurate.  The problem is people assume it to mean something it doesn't.  If I released some report about how my company bought mostly Ford trucks and said they spent more time in the shop than any of the other brands; but gave no other data, it would be correct data, but highly misleading.  BackBlaze is purposely doing this because the first time they released these numbers was simply as a cool data point for the IT community and PR reasons.  Then it exploded into a huge controversy and got them tons of free advertising, so they made some fancy graphs and they keep doing it because it keeps exploding.

 

Key things here; it is NOT a study, or a test, or anything like that.  It is simply presented that way to create hype and controversy to spread their name. 

[Doobeedoo]

Yes, repeatedly.

These idiots pulled 2.5" external drives from their enclosures and ran them 24/7 in a server environment at one point, and published the "data" as if it wasn't worthless.

 

THEY ARE MUPPETS

 

Here's some real, verified data: (Failure rates within warranty periods through French online retailers in 2014)

 

http://linustechtips.com/main/topic/255466-huge-list-of-failure-rates-for-all-pc-components/

 

Edit: I should point out that Seagate's 3TB drives have always been awful, I am not refuting that fact.

Yeah that seems more realistic list of percentages.

Though, I have 1TB and 3TB Seagate drive for year and a half and I've been using them for everything; transfer, packing, extracting, recording, editing, many delete, and so on. Even did stuff like while some transfer is going on or something I'd load a game from same drive and minimize that and do something third.

Never experienced any failures or hangs.

[dalekphalm]

What about Hitachi?

As others have mentioned, Hitachi is on there = HGST - Hitachi Global Storage Technologies. WDC bought them (Or more specifically, the HDD sector of Hitachi) a few years ago and rolled them into WDC as a sub-brand. They also bought all of HGST's manufacturing facilities too, so HGST drives aren't necessarily just WD with a Hitachi logo slapped onto it.

 

Haven't we already established that this is a bunch of BS?

Yes.

 

but even if they are like 10-15 dollars cheaper i would suppose that what every cost benefit you get is quickly lost and turn to the negatives when 40% of the drives fail

One thing of note, the 40% is from last year, not this year. They've actually seen a substantial drop in failures for Seagate compared to last year.

 

Plus, this graph does not go into the age of the HDD for failures. That's a huge missing data point, since all HDD's will fail more frequently with age. Ones that are 3 years old are much more likely to die compared to ones that are 6 months old. And ones that are 3 years old were likely budgeted for replacement anyway.

 

the grey bar goes to 40%

See above.

 

Yes, repeatedly.

These idiots pulled 2.5" external drives from their enclosures and ran them 24/7 in a server environment at one point, and published the "data" as if it wasn't worthless.

 

THEY ARE MUPPETS

 

Here's some real, verified data: (Failure rates within warranty periods through French online retailers in 2014)

 

http://linustechtips.com/main/topic/255466-huge-list-of-failure-rates-for-all-pc-components/

 

Edit: I should point out that Seagate's 3TB drives have always been awful, I am not refuting that fact.

Agreed.

 

Yes, BlackBlaze's stats are horseshit and I have no idea why people still believe them.

 

IIRC from a few years ago they were putting consumer drives under 24/7 enterprise workloads and even taking drives out of enclosures (the ones where you buy a hard drive already in the external enclosure) and putting them into server racks and whatnot. Basically, everything was mixed and matched and there was zero consistency and it was not scientific or accurate whatsoever.

Agreed.

 

Because HGST is an enterprise class drive, designed for 24/7 use in a large array, there is a price premium for these drives and naturally their failure rate is lower

HGST are not "Enterprise" class drives. HGST is a HDD manufacturer/brand (Now a subsidiary of WDC, but they still have all the Hitachi fabs). They made a wide range of desktop HDD's, from consumer to enterprise. They still do.

https://www.hgst.com/hard-drives

 

They dropped out of the "consumer desktop HDD" category, but they still make NAS, Laptop, Enterprise, and Industry Specific (Think CCTV/AV/Security, as well as CNC or other Automation) drives.

 

NAS drives are still considered consumer drives though, so there's that in addition to the Laptop drives.

 

What people need to realize about these numbers is this:

 

The numbers themselves are not wrong, but they are invalid.

Why? Because for the simple reason that they are running Consumer, cheap-as-fuck desktop drives in a VERY HIGH LOAD Server/Enterprise environment.

 

No shit 40% of your HDD's failed when you're using consumer drives in a fucking mass storage pod.

I'm surprised that some of the drives held up NEARLY as well as they did.

 

The reason why these numbers are irrelevant, is because no one who buys consumer drives should be using them in this manner. It's irresponsibility for their IT/Purchasing departments. They're being cheap fuckers instead of investing in the proper HDD's that are properly designed, tested, and verified in those environments.

[mr moose]

The problem is not the method or the data.  The problem is people think this is some kind of study or test.  There is no testing, no studying here.  It is just essentially random numbers released by some company.  They simply state that this is what they use.  Then everyone and their mother thinks it is some kind of professional study done to determine the efficacy of HDD makers.  I will say that the way they put out the data is a bit misleading, but that is mostly for the very purpose that it causes controversy, thus spreading their name around, its basically just free publicity. 

 

The data is not flawed in and of itself.  It is 100% accurate.  The problem is people assume it to mean something it doesn't.  If I released some report about how my company bought mostly Ford trucks and said they spent more time in the shop than any of the other brands; but gave no other data, it would be correct data, but highly misleading.  BackBlaze is purposely doing this because the first time they released these numbers was simply as a cool data point for the IT community and PR reasons.  Then it exploded into a huge controversy and got them tons of free advertising, so they made some fancy graphs and they keep doing it because it keeps exploding.

 

Key things here; it is NOT a study, or a test, or anything like that.  It is simply presented that way to create hype and controversy to spread their name. 

 

Except that it is a study because the act of collecting data and analyzing the phenomenon of a subject then reporting the results is exactly that by definition. The results are flawed because the arrangements and use of the drives are not clearly reported.   As I said earlier the WD drives they had could have been in extremely light use while the Seagates could have been used for the highest workload.  We just don't know, and unless all that information is made available it contradicts all the other information that is out there.  Thus a flawed study.

 

If the data is accurate, then they are going to hope they can prove that should seagate ever decide to go them for defamation.   Because after 3 annual reports all looking the same and so many people claiming their results are "myth" and "flawed" they may have a case.

[TheMcSame]

Except that it is a study because the act of collecting data and analyzing the phenomenon of a subject then reporting the results is exactly that by definition. The results are flawed because the arrangements and use of the drives are not clearly reported.   As I said earlier the WD drives they had could have been in extremely light use while the Seagates could have been used for the highest workload.  We just don't know, and unless all that information is made available it contradicts all the other information that is out there.  Thus a flawed study.

 

If the data is accurate, then they are going to hope they can prove that should seagate ever decide to go them for defamation.   Because after 3 annual reports all looking the same and so many people claiming their results are "myth" and "flawed" they may have a case.

If think you're missing the point. They're just publishing which drives have failed in their data centre, if that many drives have failed then they're accurate. They aren't putting this out as the end all be test that you seem to think they're doing.

[Rheinwasser]

They are still different drives, even though it is owed by WD, Hitachi still designs the drives. 

that sounds logical, but i still ask myself - from a noob perspective - why they are not just asking them how to make their drivers better.

[mr moose]

If think you're missing the point. They're just publishing which drives have failed in their data centre, if that many drives have failed then they're accurate. They aren't putting this out as the end all be test that you seem to think they're doing.

But they are putting it out there as a report on "what hard drive is best?",  a report on hard drive failures. 

 

https://www.backblaze.com/blog/best-hard-drive/

 

It's even in their URL which comes up as the lead title in a google search.

[TheMcSame]

But they are putting it out there as a report on "what hard drive is best?",  a report on hard drive failures. 

 

https://www.backblaze.com/blog/best-hard-drive/

 

It's even in their URL which comes up as the lead title in a google search.

Clickbait articles are a thing...

 

They're looking at which brand of hard drives are the best (or most reliable) in their data centre, anyone that even takes this seriously for consumer level use is an idiot. All they're showing is which drives have failed, they aren't claiming X is better than Y for consumer level builds.

[mr moose]

Clickbait articles are a thing...

 

They're looking at which brand of hard drives are the best (or most reliable) in their data centre, anyone that even takes this seriously for consumer level use is an idiot. All they're showing is which drives have failed, they aren't claiming X is better than Y for consumer level builds.

 

Clickbait indeed,  doesn't change the fact they are representing this data to the public as a report on which drives are the best.

[TheMcSame]

Clickbait indeed,  doesn't change the fact they are representing this data to the public as a report on which drives are the best.

Ties into the clickbait thing really, present it in a way which generates controversy and word will get around. Free publicity.

[mr moose]

Ties into the clickbait thing really, present it in a way which generates controversy and word will get around. Free publicity.

 

So you admit they are presenting this as legitimate results as to which is the best hdd to buy?

[TheMcSame]

So you admit they are presenting this as legitimate results as to which is the best hdd to buy?

It's presented that way in the graphs but it doesn't make the results inaccurate as all they're showing is how many drives have failed. Again, you'd have to be stupid to take these results seriously for consumer use.

[dalekphalm]

If think you're missing the point. They're just publishing which drives have failed in their data centre, if that many drives have failed then they're accurate. They aren't putting this out as the end all be test that you seem to think they're doing.

The "data" might be accurate, but Context Matters.

 

How heavily were the drives used? How old were they used? What is the average "Hours on per month" for each brand? What if all their WD drives are less than 1 year old and were used as monthly backup drives, while all the Seagate Drivess are 4 years old and used as scratch drives with constant, daily, non-stop usage?

 

That might be an extreme example, but the point still stands. This data is useless because it doesn't give us a breakdown of age, usage, hours on, number of TB written per anum, etc.

[mr moose]

It's presented that way in the graphs but it doesn't make the results inaccurate as all they're showing is how many drives have failed. Again, you'd have to be stupid to take these results seriously for consumer use.

 

Given some of this forum (of tech enthusiasts) take it as gospel, it's not hard to imagine that your average consumer or amateur would easily take them results as overall bona fide god's own truth.    It appears many people here don't understand statistical analysis and the importance of eliminating/reporting bias conditions.

[TheMcSame]

The "data" might be accurate, but Context Matters.

 

How heavily were the drives used? How old were they used? What is the average "Hours on per month" for each brand? What if all their WD drives are less than 1 year old and were used as monthly backup drives, while all the Seagate Drivess are 4 years old and used as scratch drives with constant, daily, non-stop usage?

 

That might be an extreme example, but the point still stands. This data is useless because it doesn't give us a breakdown of age, usage, hours on, number of TB written per anum, etc.

Context is irrelevant if you ask me, backblaze isn't out to prove anything, people are just seeing something on the internet and believing it instantly.

If backblaze was trying to prove something then sure, context would matter, but in this senario it really doesn't matter. All they're showing is how many drives have failed in their data centre.

Given some of this forum (of tech enthusiasts) take it as gospel, it's not hard to imagine that your average consumer or amateur would easily take them results as overall bona fide god's own truth.    It appears many people here don't understand statistical analysis and the importance of eliminating/reporting bias conditions.

Let idiots be idiots, that's the way I go. It doesn't take long to figure out that a data centre is going to see far more drive usage than any consumer will. Not to mention that you shouldn't just go by one source when researching parts.       

[aj101202]

Never had any issues with any of my seagate drives the 1TB and 500gb ones in my gaming pc are about 5 years old now still going fine only drive i've had fail is a WD laptop drive.

[dalekphalm]

Context is irrelevant if you ask me, backblaze isn't out to prove anything, people are just seeing something on the internet and believing it instantly.

If backblaze was trying to prove something then sure, context would matter, but in this senario it really doesn't matter. All they're showing is how many drives have failed in their data centre.

Let idiots be idiots, that's the way I go. It doesn't take long to figure out that a data centre is going to see far more drive usage than any consumer will. Not to mention that you shouldn't just go by one source when researching parts.       

How can you possibly say context is irrelevant? That's the most ridiculous statement for technology I've ever hear.

 

Wow. Did you see the pieces of information I listed in my last post? Those are incredibly important.

 

If I buy a HDD, use it once every 6 months, where it otherwise sits in a dry, cool, box, and it lasts 17 years, and then I buy the competitors equivalent, and run it 24/7 as a Video Editing scratch drive that gets worked on around the clock by a rotating shift of editors, and it lasts 1.2 years, and then just graph "Model vs failure rate" after 2 years (yet not state how long the time has been), then those numbers are 100%, completely, unquestionably pointless.

 

That doesn't tell me which drive is more likely to fail. That doesn't state ANYTHING. That takes the data and misrepresents it to the nth degree.

 

Blackblaze might not be out to "prove" anything, but by not providing all the details, they are putting out data that could very well massively misrepresent the products.

[TheMcSame]

How can you possibly say context is irrelevant? That's the most ridiculous statement for technology I've ever hear.

 

Wow. Did you see the pieces of information I listed in my last post? Those are incredibly important.

 

If I buy a HDD, use it once every 6 months, where it otherwise sits in a dry, cool, box, and it lasts 17 years, and then I buy the competitors equivalent, and run it 24/7 as a Video Editing scratch drive that gets worked on around the clock by a rotating shift of editors, and it lasts 1.2 years, and then just graph "Model vs failure rate" after 2 years (yet not state how long the time has been), then those numbers are 100%, completely, unquestionably pointless.

 

That doesn't tell me which drive is more likely to fail. That doesn't state ANYTHING. That takes the data and misrepresents it to the nth degree.

 

Blackblaze might not be out to "prove" anything, but by not providing all the details, they are putting out data that could very well massively misrepresent the products.

If you want to give more ammo to fanboys then more power to you, I'm just saying that if they aren't out to prove anything what use is it mentioning what these drives are being used for?

 

Either way, you guys are taking this far too seriously and almost come across as damage control to me...