Backblaze Q2 Reliability Results

S

Sp33dFr33k

2[H]4U
Joined
Apr 20, 2002
Messages
2,481
Each quarter, Backblaze updates our hard drive statistics with the latest data. As of the end of Q2 2015, there were 47,561 drives spinning in our datacenter. Subtracting out boot drives, drive models with less than 45 drives and drives in testing systems, we are publishing data on 46,038 hard drives spread across 21 different models, from 1.5TB to 8.0TB in size.

https://www.backblaze.com/blog/hard-drive-reliability-stats-for-q2-2015/
 
I am most interested in the Tosbiba results being that the 5TB model goes on sale for $139.
 
Thank you for sharing. I follow these updates on Hacker News too.

The Seagate results supplement my feelings toward their drives. The only Seagate I've used died w/in a few months. Returned it for another and that one died in about the same amount of time. I've never purchased Seagate since, but hope others have had better luck with them.
 
WDC Red is not looking good with that either however they have a much smaller sample than Seagate or HGST.
 
drescherjm said:
I am most interested in the Tosbiba results being that the 5TB model goes on sale for $139.
Click to expand...

Yeah.. i've been looking 4x 5TB drives as a storage upgrade for my server box. Been looking at WD Reds but those are like ~$70 more per drive than the Toshiba 5TB drives. (spindle speed doesn't matter to me, these are low usage bulk storage drives)
 
interesting toshibas seem to be decent with 4-5% failure rate. They beat the Reds at least. Those HGSTs are pretty damn good. Glade I got toshibas since they are cheap as dirt and reliable for the most part
 
Looks like Seagate got something right with the ST4000DX000, for a change. I wonder if it's dumb luck or are they growing a brain over there. :D
 
drescherjm said:
WDC Red is not looking good with that either however they have a much smaller sample than Seagate or HGST.
Click to expand...

Considering that all 8 of the hard drives in my file server are the WD Red 3TB, I'm not particularly happy seeing those results. With that said, I haven't had that much bad luck with the Seagate 3TB drives so...*knock on wood* let's see what happens.

This is why you keep backups, people!
 
Twenty 2TB HGST deskstars in my primary file server, and eight seagate 3TB HD's in my backup file server.

I mean.... everything is backed up, but... those figures for the 3TB Seagates has me very concerned now -__-
 
iamwhoiamtoday said:
Twenty 2TB HGST deskstars in my primary file server, and eight seagate 3TB HD's in my backup file server.

I mean.... everything is backed up, but... those figures for the 3TB Seagates has me very concerned now -__-
Click to expand...

48 drives or whatever is a horribly small sample. It is said a good sample is 1250 IIRC. at least the toshibas are 146 or something in sample size
 
rive22 said:
Looks like Seagate got something right with the ST4000DX000, for a change. I wonder if it's dumb luck or are they growing a brain over there. :D
Click to expand...

Yes but those are the slower drives. I know seagate made a big deal out of this by getting rid of their green and performance drives.. Now they find they can not have performance AND reliability at the same time without driving up costs. So I wonder how they are going to get out of this bind.. Ofcourse since marketing can make all kinds of wonder blunders not knowing the technological challenges, you cant talk out of both sides of your mouth. Considering seagate drives are not much cheaper than higher performance drives and reliability is the same, I will stick with others. Knowing seagate, they will start to talk from the other side as soon as they can source cheaper parts that can perform better and then charge more. but due to all the gimmicks, you can never reliably take the specs at face value as they dont publish it anymore. If I send in a drive for replacement, I expect to get an identical one back. Not an inferior one! Especially if they were only a months old.
 
SomeGuy133 said:
48 drives or whatever is a horribly small sample. It is said a good sample is 1250 IIRC. at least the toshibas are 146 or something in sample size
Click to expand...

It is enough to get a sample reading on their reliability. If you have many failures then they have a major problem with component reliability. Even if you get 10K drives, if they are ALL from the same batch, made at the same time at the same factory, either they ALL will develop problems and die soon or they all will not have any problems. As all the components will show similar tolerance just due to QA. The problems start appearing when they start to change components and start building them at different plants under different conditions and such, we can only get such info when sourcing drives in small quantities over a period of time. Which would show the QA of the company as a whole rather than just a single plant... As we know, some plants are run by competent people and some are run by those who show up for work only because they have to. This is why when a new plant is opened their output usually have some real funky reliability problems. Even Toshiba had major problems with their plant in the Philippines and reliability were poor. So it can not just be the components but the people setting things up and running things. That same plant seems to have increased reliability by a wide margin in a few years as they got rid of the unreliable elements. Seeing HGST's 1-2% failure rates over multiple years I got to say they must have some very talented people working there over the long term. BAD untrained workers will cause all kinds of problems with high tech manufacturing no matter how closed their product cycles are.
 
HGST is definitely a slam dunk. They make ya pay for it though :cool:
 
As long as a drive is under 10% I would consider it good enough. If a drive has considerable life expectancy past 1 year I will consider it. Seeing as most of these drives (minues a few select Seagates) have great life and most are under 10% failure rate, of these should be considered for personal use.
 
iamwhoiamtoday said:
Twenty 2TB HGST deskstars in my primary file server, and eight seagate 3TB HD's in my backup file server.

I mean.... everything is backed up, but... those figures for the 3TB Seagates has me very concerned now -__-
Click to expand...

Read the whole thing first. Yes, the Seagates have a REALLY bad rate...Then read at the bottom, they state the Seagates are the oldest drives by far in their data-center, average age for the Seagates is over 5 years, so they have been in service a LONG time, while all the other drives are much newer, so having a much higher failure rate on drives that have upwards of 3 extra years in service is not shocking. At 5+ years for these drives in a server environment and server class usage, I don't think that is to bad. Some of the other brands, such as the newest WD drives, they had as high as a 13% failure rate in just 8 months, they said over time that had gone down, but this is a year by year report, not a total life of each type of drive, had they done this, the Seagates would have a much much lower failure rate.

Granted, I have not had a Seagate drive in a long ass time, however the last ones I had served me well, but I have been almost all WD and HGST, with most of the newer drives being Toshiba (5TB). No problems with them so far in the short time and very small sample size I have, the WD have been great, only one failed, but it was well over 6 years old in a computer thats on 24/7, granted not server like use that these drives see, the HGST however have never failed, but I don't use anymore, they are, gosh...7-8 (or more maybe) years old and as such all no bigger than 300GB, which was really large for the time, they just took up more space than they were worth and got changed out.
 
what they need to do is make a cumulative line graph showing failures and age -_- That is what would be most meaningful :/
 
SomeGuy133 said:
what they need to do is make a cumulative line graph showing failures and age -_- That is what would be most meaningful :/
Click to expand...

Indeed, but unlike allot of other places they do give all the raw data for you to DL and you could make your own, or at least get an idea based on age and failure rates over time.
 
BlueFireIce said:
Indeed, but unlike allot of other places they do give all the raw data for you to DL and you could make your own, or at least get an idea based on age and failure rates over time.
Click to expand...

yes but the extra 5 mins it takes them saves 100 or more people 10 mins each -_- Stupid not to but at least they publish it compared to every other place in the world :/
 
I am worried about those Toshiba results. I bought 16 of the 5TB drives when they were last on sale.
 
Chellexelle said:
I am worried about those Toshiba results. I bought 16 of the 5TB drives when they were last on sale.
Click to expand...

Why?

That is an acceptable rate for me for home use, also keep in mind that the WD drives, which everyone loves they stated had as high as a 13% failure rate. The year that the drives first get added are going to have a high rate because of bad drives and all the ones that fail in the first month or so, after a few months all those drives are gone and you are left with mostly good drives. After this the drives tend to have a much similar failure rate which then changes years into service when you start to see the normal long term life span of the drives in question.
 
BlueFireIce said:
Read the whole thing first. Yes, the Seagates have a REALLY bad rate...Then read at the bottom, they state the Seagates are the oldest drives by far in their data-center, average age for the Seagates is over 5 years,
Click to expand...

Actually, the stat is for failures PER YEAR.
 
atom said:
Actually, the stat is for failures PER YEAR.
Click to expand...

The main stats everyone is talking about are on the main page which are listed per quarter, they don't give all the data, and the main page only shows back to 4Q 2013, where you can see many of the other dives were not even in service back then, while the Seagate were in service back in 2009 and earlier.

So nothing I said is incorrect.

You can also see that the newer Seagates that went into service the same time the WD and HGST drives have just as good of a failure rate, and better than almost all the WD drives. Comparing old/last gen drives to newer drives from someone else is apples to oranges, as well, the older drives have been in service for many more years, making the numbers even further skewed.
 
But if you take a basket of fruit and you find the majority of them contain worms, I sure don't want to find the ones without them even if they are cheaper.. I would rather just find baskets from companies thats clean and without chemicals covering them to keep them fresh.

Hard to argue with statistics and trends.. But keep going, you can make a point for and against using the same numbers.. Like Toshiba 4TB drives are the worst..
 
AlienTech said:
But if you take a basket of fruit and you find the majority of them contain worms, I sure don't want to find the ones without them even if they are cheaper.. I would rather just find baskets from companies thats clean and without chemicals covering them to keep them fresh.

Hard to argue with statistics and trends.. But keep going, you can make a point for and against using the same numbers.. Like Toshiba 4TB drives are the worst..
Click to expand...

No.

It is more like saying, this fresh basket of fruit has almost no worms and taste great, but this other guy, who has baskets that are weeks old that he is about to throw out have worms! Just don't look at the new baskets he is selling that have even less worms than mine, those don't count!!!

:rolleyes:

The data is right there, I don't know what is so hard to get about that? Is fanboyism that strong or is it the hate for Seagate that is that strong? I am not even saying Seagate is good or better, just saying that the numbers are there and the new Seagates have orders of magnitude better failure rates than the new WD.

I use to have Seagate, at the time they had good track records and good price, then those were getting replaced with WD and Hitachi drives as at the time I was buying they had the better price and good record, right now for my 5TB drives (Toshiba), they have a large price advantage and have a good enough track record for home use (received by 4th the other day). If Seagate had the size drives I wanted at a better price and the record the newer drives have, I would have gone with them.
 
The statistics tell me that the newer Seagates are fine and Hitachi are clearly the best. Other than that the WDC Red and Toshiba drives do not look good but they are very small samples.
 
You must log in or register to reply here.
Back
Top