Backblaze Hard Drive Stats for Q2 2016

FrgMstr

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,634
Now a little something for all you storage nerds out there. We all know the favorite discussion among us is what drives suck the most, or least, when it comes to storage. Considering the nerds at Backblaze had over 68,000 drives installed in Q2, I would suggest they are as qualified as any of us to talk about hard drive failure rates over time. Below is the meat and potatoes of the data, but the page in its entirety has some good data as well.
 
Have them ship you the bad ones so you can do more 50 cal vs hdd videos. Or I wonder how ssds will do vs a 50 cal?
Back on topic its nice to see failure rates and i wish Hardware manufacturers would put the info out themselves.
 
Two weeks ago I replaced a WD 500 GB (ran for many years, but alas succumbed to hard drive death) and a Seagate 1.5 TB. The Seagate had a decent life, may it rest in pieces. (I'm going to disassemble both so I can play with the magnets and destroy the platters).
 
I always look forward to these numbers.

Again HGST rules unchallenged, but the big news to me is the sharp turn for the worse in the WD numbers.
 
Some of those failure rates are really surprising. 9% is really, really bad.
 
Ultima99 said:
I always look forward to these numbers.

Again HGST rules unchallenged, but the big news to me is the sharp turn for the worse in the WD numbers.
Click to expand...

I used to have great luck with WD drives, but I've had 4 drives fail over the past 6 months. These are 4TB drives that are a little over 3 years old.
Failure rate is almost as bad as the Seagate drives Dell used a several years ago.
 
I'm personally shocked by these numbers! My personal experience is much different. I find WD drives to be the best followed by Seagate a long way back. I have had one HGST drive after another fail. Toshiba drives are paper weights IMO. At my previous employer, HGST SAS drives were so bad, at one point we asked the vendor who was providing the service contract to quote us replacement costs to replace every HGST drive in the SAN. For a long time Seagate made bullet proof Enterprise drives but just like their desktop brethren they started to slip too. Don't even get me started on failure rates for HGST and Toshiba laptop drives! I'm trying to think of the last WD drive that failed me during the warranty period. I can't seem to think of any. I wonder how much spin-up / spin-down factors into these failure rates...
 
I already replaced 6 3 TB sea gates in the last 3 or so years in my nas.

Moved to wd red and so far, no problems.
 
I remember them saying we would see a spike in WD numbers as they were starting to get old.
 
Likely reason they will be shifting to SSD. They see the writing on the wall as well. Yes these are pretty sad numbers though. WD black drives are still super expensive. The blue drives are a lot also. I saw a 4TB at the store for $139. The same price as seagate. No price fixing here folks.
 
Zareek said:
I'm personally shocked by these numbers! My personal experience is much different. I find WD drives to be the best followed by Seagate a long way back. I have had one HGST drive after another fail. Toshiba drives are paper weights IMO. At my previous employer, HGST SAS drives were so bad, at one point we asked the vendor who was providing the service contract to quote us replacement costs to replace every HGST drive in the SAN. For a long time Seagate made bullet proof Enterprise drives but just like their desktop brethren they started to slip too. Don't even get me started on failure rates for HGST and Toshiba laptop drives! I'm trying to think of the last WD drive that failed me during the warranty period. I can't seem to think of any. I wonder how much spin-up / spin-down factors into these failure rates...
Click to expand...

Yeah I've used all WD drives in both my main system and my secondary system for the past 10-15 years and the last one that failed was a 60GB "Blue" PATA/IDE drive it's been so long. Most of those drives ran for 5+ years too, since they get moved to secondary PC when my main PC is done with them. I'm sure I use them pretty lightly compared to Backblaze though, mostly just DVD/Bluray rips.

Laptop drives in general fail a lot (unless they are SSDs) since that is sort of a nightmare environment for a hard drive, although even in that case WD seems to be a bit better just judging based on people's laptops I have worked on.

All that being said you are fool if you expect any drive to be 100% reliable. If you do be prepared to lose a lot of data at some point.
 
Backblaze numbers should be taken with a metric ton of salt. Their use case is nothing like what most of the drives are designed for.

They use consumer drives, not enterprise drives, for data center purposes. The enclosures are basically drives wrapped in a rubber band and stacked directly next to each other. Consumer drives are not meant to take that vibration and heat. In fact, stacking the drives that close together, all on the same axis, creates very bad amplification effect.

It's like taking a Toyota Camry, one of the most reliable cars on the road, on the Rubicon trail and wondering why it breaks down.

These failure rates are way higher than any of these drives would see in normal PC use.
 
I've been sticking to WD black drives as my main storage device for files other than the OS being on an SSD and (crosses fingers) have yet to have an issue. I have however just recently needed to replace a hitachi drive in my wifes PC for having snowballing bad sectors.
 
tgom222 said:
Backblaze numbers should be taken with a metric ton of salt. Their use case is nothing like what most of the drives are designed for.

They use consumer drives, not enterprise drives, for data center purposes. The enclosures are basically drives wrapped in a rubber band and stacked directly next to each other. Consumer drives are not meant to take that vibration and heat. In fact, stacking the drives that close together, all on the same axis, creates very bad amplification effect.

It's like taking a Toyota Camry, one of the most reliable cars on the road, on the Rubicon trail and wondering why it breaks down.

These failure rates are way higher than any of these drives would see in normal PC use.
Click to expand...

Yes, but if you take a Toyota Camry, one of the most reliable cars on the road, on the Rubicon trail next to a Honda accord and 10 other similar classed vehicles And see who lasts the longest, you have a pretty good chance the more reliable road cars will ALSO be the more reliable off-road cars.

The point is that there is no "Off Road" for hard drives. You either use them a lot, or don't use them much. You either keep them in a tight space, or you don't. Better drives tolerate more usage and abuse, these are VERY few situations wherein one drive that is rated for datacentre use will somehow last LESS time in a NAS than a 'NAS rated' drive in the same NAS. The Datacentre drive is better on all accounts, in all situations and use-cases: it lasts longer because its designed to last longer. There is no "Off Road" hard drive: Better drives last longer and put up with more abuse. Yeah, these drives are put in pretty terrible circumstances, but these circumstances equal across the board, in other words: its fair to each drive. Its not like the only put the seagates through hell and keep the WDs in a water cooled enclosure, being fanned by palm leaves. All drives experience the same circumstances.
 
KazeoHin said:
Yes, but if you take a Toyota Camry, one of the most reliable cars on the road, on the Rubicon trail next to a Honda accord and 10 other similar classed vehicles And see who lasts the longest, you have a pretty good chance the more reliable road cars will ALSO be the more reliable off-road cars..
Click to expand...


BINGO!

If I'm forced into expanding my storage on spinning disk, I know what brand it'll be. I know exactly what part number it'll be!
 
I'm suprised the WD Reds were not doing so good. Makes me sad for my NAS.
 
Replaced 22 3TB Seagate ST3000 POS's in my 3 home NAS's with 22 4TB HGST's... been very happy with 0 drive failures in ~2 years.

My seagates were falling off at a rate of ~2-3 a month. What a tremendous waste of money on that garbage.

Just bought 10 more 2TB HGST's to replace 8 more drives in my parents 8 bay NAS after I had 4 Seagates take a dump all at the same time. Those Seagates were SV35/ST2000's.

I'm not having a good track record with Seagate...
 
ChrisJohnson00TA said:
Replaced 22 3TB Seagate ST3000 POS's in my 3 home NAS's with 22 4TB HGST's... been very happy with 0 drive failures in ~2 years.

My seagates were falling off at a rate of ~2-3 a month. What a tremendous waste of money on that garbage.

Just bought 10 more 2TB HGST's to replace 8 more drives in my parents 8 bay NAS after I had 4 Seagates take a dump all at the same time. Those Seagates were SV35/ST2000's.

I'm not having a good track record with Seagate...
Click to expand...
Jeez man, how much 8k porn do you need?!
 
Damn... I just bought 4 x 4TB WD Reds yesterday for $129 each for my ESXi server, pondering if I should return them. Would've gotten the HGST but I was bit concerned about the noise from a 7200 RPM drive.
 
But in that article they said they use Seagates a lot. Only certain Seagates were bad. I had 2 WD 2TB Blacks go bad on me so replaced with a Seagate ST2000DM001-1ER1.

"Which hard drives do we use?

We’ve written previously about our difficulties in getting drives from Toshiba and Western Digital. Whether it’s poor availability or an unexplained desire not to sell us drives, we don’t have many drives from either manufacturer. So we use a lot of Seagate drives and they are doing the job very nicely."
 
Last edited:
rezerekted said:
But in that article they said they use Seagates a lot. Only certain Seagates were bad. I had 2 WD 2TB Blacks go bad on me so replaced with a Seagate ST2000DM001-1ER1.

"Which hard drives do we use?

We’ve written previously about our difficulties in getting drives from Toshiba and Western Digital. Whether it’s poor availability or an unexplained desire not to sell us drives, we don’t have many drives from either manufacturer. So we use a lot of Seagate drives and they are doing the job very nicely."
Click to expand...

Look at the CONFIDENCE INTERVAL. What matters is not the number of drives...but in reality having enough failures coupled with the amount of drives to make the probability of failure accuracy "good enough to make an informed/educated decision". If you look at Seagate compared to HGST there is NO question; Seagate is worse. Now whether or not the price difference in drives makes up for the cost of labor to swap the drives is another story. If you look at the size class it becomes even more evident. The only outlier is how good SG 6TB performs.
 
KazeoHin said:
Yes, but if you take a Toyota Camry, one of the most reliable cars on the road, on the Rubicon trail next to a Honda accord and 10 other similar classed vehicles And see who lasts the longest, you have a pretty good chance the more reliable road cars will ALSO be the more reliable off-road cars.

The point is that there is no "Off Road" for hard drives. You either use them a lot, or don't use them much. You either keep them in a tight space, or you don't. Better drives tolerate more usage and abuse, these are VERY few situations wherein one drive that is rated for datacentre use will somehow last LESS time in a NAS than a 'NAS rated' drive in the same NAS. The Datacentre drive is better on all accounts, in all situations and use-cases: it lasts longer because its designed to last longer. There is no "Off Road" hard drive: Better drives last longer and put up with more abuse. Yeah, these drives are put in pretty terrible circumstances, but these circumstances equal across the board, in other words: its fair to each drive. Its not like the only put the seagates through hell and keep the WDs in a water cooled enclosure, being fanned by palm leaves. All drives experience the same circumstances.
Click to expand...


If things like noise and price are equal then sure, might as well prepare for the apocalypse. Otherwise, you are taking on additional cost and potentially noise, to fend off vibration and heat issues that won't exist in a home-usage scenario.
 
brentsg said:
If things like noise and price are equal then sure, might as well prepare for the apocalypse. Otherwise, you are taking on additional cost and potentially noise, to fend off vibration and heat issues that won't exist in a home-usage scenario.
Click to expand...

Except home use scenarios aren't free of vibration, they certainly have less of it, but it isn't non-existent. If noise and price are not a concern for a home use scenario, then it's not an issue. The information is simply more data that a user can take into consideration when making their HDD purchase decisions, along with other factors like cost and noise.

Also, not every home usage scenario involves a singular HDD sitting in a desktop. Hell, not every person interested in the information presented is going to be using it as a factor in purchase decisions for home usage.
 
Ihaveworms said:
Jeez man, how much 8k porn do you need?!
Click to expand...

You can never have enough 8k porn...

I have one NAS mirrored to the other... 8 drives in raid 6 - ~22TB usable, I'm hovering around 18tb used. Obviously mirrored, so x2. Funny.. no porn.

My 6 bay is used for iSCSI to my main desktop. Also raid 6 - ~15tb to my desktop - mostly used for VM's and image processing from the main rig. Currently 7tb used.

My dad uses his for torrent storage and media streaming in his house. Lets call it ~10tb and not sure on his current use.
 
Contrary to some people's experiences here, my last 3 out of the last 4 drives i bought are/were Hitachi's. The most used one is dead 2tb, the lighter use one 1,5tb (powered on for browsing and some office sometimes) has bad sectors 1,5tb, and the one that is not used is ok 1,5tb, for now at least. Aside from these 2 only had a Quantub fail back in the day. Extremly dissapointed in them to say the least. Will personally avoid them.
 
Doesn't matter if they were designed for it or not, it still a good indicator of quality of the drives.
 
I've the following drives and my experiences with them.

1 x 2TB WD USB3 Portable - Works great after 4 years
2 x 2TB WD USB2 Desktop drives - Working great after 4 years (most of the time in storage)
2 x 3TB Seagate USB3 Desktop drives - Lightly used if any. One is having problems with write, sometimes the write speed is less than 1MB/s in some areas even when copying large contiguous files; I'm afraid to test the other, but will bring it out of storage soon to replace the backup in it.
2 x 4TB Seagate USB3 Desktop drives - Both lightly used. One failed to be read, the other one is beginning to randomly disconnect whenever I copy large files... (these two are the newest ones I've had... a little more than 2 years!)

Seems like all my Seagate drives are failing...

And when I saw a 4TB Portable Seagate for $99, I bit. Doh!
 
Monkey God said:
Doesn't matter if they were designed for it or not, it still a good indicator of quality of the drives.
Click to expand...
sure it does, something about vibrations and harmonics, heat/cool cycles, electrical noise from a large number of drives on the same power rails.

so many possible variables that most consumer computers never see since they are single drives including the sheer read and write rates...
 
omegatotal said:
sure it does, something about vibrations and harmonics, heat/cool cycles, electrical noise from a large number of drives on the same power rails.

so many possible variables that most consumer computers never see since they are single drives including the sheer read and write rates...
Click to expand...

They see all the same variables just not to the same extreme.
 
You must log in or register to reply here.
Back
Top