Drive reliability for NAS... now questioning my original choice

[copenhagen69]

After all the help I received in the last thread on server/cloud storage solutions I decided to with the NAS option.

I was pretty sure I wanted to go 3TB WD Red drives for my NAS but then I found some posts that made me question that decision ...

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

Western Digital Red 3 TB (WDC WD30EFRX) 3TB 0.9 years life span and 12.87% failure rate annually ...

After I saw that it made me instantly question my choice.

From looking around I have found that pretty much this is what people seem to agree on ...
Always Hitachi if price is no object, Seagate because the bang-for-the-buck of the 4TB drives is hard to beat (but keep an eye on them, especially as they get older), and WD Reds for the best overall balance of price/reliability.



thoughts?

 

[rtwjunkie]

Can't speak to the 3TB ones, but I have 6 of the 2TB Reds in my server, 4 of which are well into their second year (approaching 2 full years), and not one sign of being any worse for the wear. The other two are 6 months old and still fantastic as well. SMART reports great on all of them, and temps are fine. Read/Write as quick and responsive as ever.

 

[Sasqui]

Here's an article refuting Backblazes methods and analysis: http://www.tweaktown.com/articles/6...bility-myth-the-real-story-covered/index.html

That said, I'd shell out more for the WD Black series. In my opinion, the 5 year warranty speaks for itself. My own experience with WD blacks has been nothing short of amazing, but my own singular experience is anecdotal.

 

[rtwjunkie]

Here's an article refuting Backblazes methods and analysis: http://www.tweaktown.com/articles/6...bility-myth-the-real-story-covered/index.html

That said, I'd shell out more for the WD Black series. In my opinion, the 5 year warranty speaks for itself. My own experience with WD blacks has been nothing short of amazing, but my own singular experience is anecdotal.

I trust my own anecdotal experience as well as Tweaktown much more than BackBlaze!

 

[copenhagen69]

Here's an article refuting Backblazes methods and analysis: http://www.tweaktown.com/articles/6...bility-myth-the-real-story-covered/index.html

That said, I'd shell out more for the WD Black series. In my opinion, the 5 year warranty speaks for itself. My own experience with WD blacks has been nothing short of amazing, but my own singular experience is anecdotal.

oh that is a good read there! I guess I wont ever be going back to that site again .. thanks!

hmmm I never thought of the blacks for NAS use before ... they have been great drives for my desktop though and really enjoyed them in Raid0 ... I am curious if the price jump on them is worth it for a NAS that wont need the extremely fast speeds for some photos/vids and movie watching. I wouldnt want to waste their ability just sitting there and could put the money elsewhere, like more RAM in the NAS.

 

[Kursah]

I've had more WD Black series fail than any other drive, but that 5 year warranty does speak for itself. You take a risk with any modern platter drive. Honestly, as many recommend, get different drives of the same capacity. That way if you end up getting a bad drive out of a bad batch, not all of your drives came from that batch. I try to do that where possible. Honestly I've had great luck, but I don't pay the premium for Black drives anymore...RMA takes forever, and my last 2 RMA replacements had bad sectors or failures within months. So maybe they NEED the better warranty for the hotter running, performance-oriented drives?

 

[rtwjunkie]

And then there is me. I've had great luck with WD Blacks. In fact, the only reason I replace is because I exceed their capacity. Have 4 in use in 2 computers as game hard drives, and one as a data storage drive in an external enclosure for my server.

 

[Sasqui]

And then there is me. I've had great luck with WD Blacks. In fact, the only reason I replace is because I exceed their capacity. Have 4 in use in 2 computers as game hard drives, and one as a data storage drive in an external enclosure for my server.

Look at my sig I've never had a single WD fail on me... and I even have two more 1TB WD blacks in my NAS

 

[Sasqui]

oh that is a good read there! I guess I wont ever be going back to that site again .. thanks!

hmmm I never thought of the blacks for NAS use before ... they have been great drives for my desktop though and really enjoyed them in Raid0 ... I am curious if the price jump on them is worth it for a NAS that wont need the extremely fast speeds for some photos/vids and movie watching. I wouldnt want to waste their ability just sitting there and could put the money elsewhere, like more RAM in the NAS.

Just me... rather have a reliable hard drive than more space or ram. I thought WD also makes and enterprise version of the red drives with 5-year coverage?

 

[rtwjunkie]

Just me... rather have a reliable hard drive than more space or ram. I thought WD also makes and enterprise version of the red drives with 5-year coverage?

Yes they do, the Red Pro. Also, the RE series are basically Enterprise versions of the Blacks. Rugged as hell, I've owned a few and were very worth it.

 

[copenhagen69]

Just me... rather have a reliable hard drive than more space or ram. I thought WD also makes and enterprise version of the red drives with 5-year coverage?

Yes they do. The Pros but they say they are for 8-16 HDD NAS systems. I'm not sure if that messes anything up when you only have 4?

From there site ...

WD Red Pro NAS Storage NAS Hard Drive
The next generation of hard drives for medium to large business network attached storage systems.

Storage for 8 to 16 bay NAS solutions
Joining the original color of NAS, WD Red Pro continues the formula of success that has led the WD Red product family by adding support beyond consumer, SOHO, and small business markets into medium and large 8-16 bay business storage systems. WD Red Pro hard drives integrate WD's exclusive technology, NASware 3.0, to provide unparalleled support for drive compatibility, reliability, and performance.

Professional solutions
Designed specifically with SMB customers in mind the WD Red family has expanded with the addition of WD Red Pro, which is available for the 8 to 16 bay medium to large-scale NAS environments. With capacities up to 4 TB, WD Red Pro carries a 5-year limited warranty and was engineered to handle the increased workloads from your business.

 

[rtwjunkie]

@copenhagen69 no I don't think it messes anything up. I think they are for 8-16 because they are more resistant to vibration and heat, vibration especially, that you get in large hdd setups.

 

[copenhagen69]

That does make sense ... I also found this review, which does make sense on why they warranty certain things the way they do ... Posted about the WD Red Pros

These are premium NASware 3 drives (newest version) and are made for use in any type NAS. Notice that these drives have a 5 year warranty. The more drives in an array, the closer the tolerance between drives becomes more important. You could use this drive as an only drive but you would not need the close tolerance specs. I have 5 Raid Enclosures in use, each with 4 drives. I have off the shelf consumer drives in some of them but most have been upgraded to premium drives because of block failures. The drives with block failures will rebuild themselves more often. The info on these drives on Amazon is quite good and accurate.

Bottom line is that the more drives in an array, the closer is the required tolerance between individual drives for 24/7 operation.

I've been in the computer business for many years and I still think that the manufacturers test results determine which label they get and the warranty level. If the drive passes the highest technical spec, then it is a premium drive. If it doesn't it gets a different label, price and warranty. The bottom of the heap gets a shorter warranty and low price.

I came out of an industry where the test results of solid state devices determined the warranty and price. I had a discussion with a Seagate Engineer and he said that if they guarantee a drive for 5 years, they only expect to have a very small percentage fail during the warranty.

Sorry for being so wordy but selecting drives is very important! After all we want our data to remain un-corrupted. In 10 years I've never failed to recover data on Raid 5 arrays. Yes, there were times when I was really sweating the recovery. I hope this info helps.

 

[claes]

Here's an article refuting Backblazes methods and analysis: http://www.tweaktown.com/articles/6...bility-myth-the-real-story-covered/index.html

This article is really, really bad and doesn't take into account:
1. BackBlaze's own caveats about their data - the three faults TT points to are clearly accounted for in BB's study (sourcing, enclosures, and environment).
2. BackBlaze's actual data - if you look to the study you'll find that TT's claims are grossly exaggerated.
2a. Most drives are bought OEM, which isn't ideal, but is not "most drives are pulled from enclosures"
2b. Even if some are pulled from enclosures or bought from shady vendors, BB doesn't make broad claims about manufacturers, they make claims about specific model numbers.
2c1. BB concedes that there environment is not like most home-user environments (this should be obvious) - but, if you look into HW return/failure rates (there's a french distributor that keeps track of these numbers internationally) you'll notice that the numbers match up surprisingly well.
2c2. More importantly, TT is flat out wrong about the conditions being worse - modern HDDs lift their heads when vibration limits are exceeded and actually perform better and last longer in "hotter" conditions (~45*C) than in cooler conditions. Sure, BB's storage pods are definitely not representative of a typical home user with 2 or 3 HDDs, but the disks are (mostly) designed to take these circumstances into account ("mostly" because some drives have really bad parking for 24/7 use).
3. Google's confirmation of these studies - Google has confirmed BB's numbers year after year, and BB also updates their study every year.

I trust my own anecdotal experience as well as Tweaktown much more than BackBlaze!

You really shouldn't trust tweak town - look at their reviews! They are a pr firm at best.

oh that is a good read there! I guess I wont ever be going back to that site again .. thanks!

hmmm I never thought of the blacks for NAS use before ... they have been great drives for my desktop though and really enjoyed them in Raid0 ... I am curious if the price jump on them is worth it for a NAS that wont need the extremely fast speeds for some photos/vids and movie watching. I wouldnt want to waste their ability just sitting there and could put the money elsewhere, like more RAM in the NAS.

Blacks are great drives, but they don't hold up in NAS situations as well as slower spinning drives with power saving features (although you can edit a Black's FW to adjust to your needs).

My experience: Have purchased ~3,000 HDDs for various RAID installs for customers, monitored over three years now. My numbers, per model, tend to match up w/ BackBlaze's.

 

[rtwjunkie]

@claes I said "more than", not that I trust them completely. Backblaze takes the absolute worst situation drives could be in and then includes loads of cheap, consumer drives in there as well, and then tries to apply their reliability results to brands. No thanks, I don't trust them.

 

[copenhagen69]

This article is really, really bad and doesn't take into account:
1. BackBlaze's own caveats about their data - the three faults TT points to are clearly accounted for in BB's study (sourcing, enclosures, and environment).
2. BackBlaze's actual data - if you look to the study you'll find that TT's claims are grossly exaggerated.
2a. Most drives are bought OEM, which isn't ideal, but is not "most drives are pulled from enclosures"
2b. Even if some are pulled from enclosures or bought from shady vendors, BB doesn't make broad claims about manufacturers, they make claims about specific model numbers.
2c1. BB concedes that there environment is not like most home-user environments (this should be obvious) - but, if you look into HW return/failure rates (there's a french distributor that keeps track of these numbers internationally) you'll notice that the numbers match up surprisingly well.
2c2. More importantly, TT is flat out wrong about the conditions being worse - modern HDDs lift their heads when vibration limits are exceeded and actually perform better and last longer in "hotter" conditions (~45*C) than in cooler conditions. Sure, BB's storage pods are definitely not representative of a typical home user with 2 or 3 HDDs, but the disks are (mostly) designed to take these circumstances into account ("mostly" because some drives have really bad parking for 24/7 use).
3. Google's confirmation of these studies - Google has confirmed BB's numbers year after year, and BB also updates their study every year.


You really shouldn't trust tweak town - look at their reviews! They are a pr firm at best.

Blacks are great drives, but they don't hold up in NAS situations as well as slower spinning drives with power saving features (although you can edit a Black's FW to adjust to your needs).

My experience: Have purchased ~3,000 HDDs for various RAID installs for customers, monitored over three years now. My numbers, per model, tend to match up w/ BackBlaze's.

a lot of info in there ... just curious if you have links to the french distributor or google? I just am curious what their information says ...

 

[claes]

With respect, this is all mostly inaccurate.

Backblaze takes the absolute worst situation drives could be in

I think we can both agree that BB's storage pods are not "the absolute worst condition" drives could be in. Honestly, it's surprising they don't see much higher failure rates. That said, having purchased and built my own storage pods, I feel confident in saying that they aren't much worse than most desktop NAS solutions and better than a lot of desktop solutions (drives don't like to be held in tight, stiff, conditions - they need room to vibrate!!!).

and then includes loads of cheap, consumer drives in there as well,

I'm not sure what your point is here - how will anyone know about how cheap, consumer drives perform and their reliability if no one tests them?
Google uses loads of consumer drives for their own storage needs, and yet the internet still works (and BB is still in business doing huge backups for thousands of customers).
Moreover, if you'd read the study you'd note that they have used enterprise drives in these conditions (as has Google, and as have I) and found reliability to be about the same. They (we) buy consumer drives because all hard drives fail - it is much more cost efficient to buy cheaper consumer drives that die a little sooner than paying the premium for a few more months of life, particularly in storage applications.

and then tries to apply their reliability results to brands. No thanks, I don't trust them.

Again, if you had read the study, you would notice that they do not do this - they specifically make claims about specific models of HDD.

For example, in their latest study, despite the 7200.11 fiasco that Seagate faced a few years ago and the meddling results of the 7200.12, and some of Seagate's drives having enormous failure rates still, BB still recommends those Seagate drives that have low failure rates. It is, quite plainly, the exact opposite of taking the results of a model and applying it to an entire brand.

No thanks, I don't trust them.

To each his own

 

[rtwjunkie]

My point about them using cheap consumer drives is because they spread FUD about what may be a perfectly usable and reliable drive for the average consumer.

 

[claes]

a lot of info in there ... just curious if you have links to the french distributor or google? I just am curious what their information says ...

No problem

http://static.googleusercontent.com/media/research.google.com/en/us/archive/disk_failures.pdf
http://www.hardware.fr/articles/934-1/taux-retour-composants-12.html - most recent I think
http://storagemojo.com/2007/02/19/googles-disk-failure-experience/ - these guys do a great job of analyzing the veracity of these studies and doing their own research
http://en.wikipedia.org/wiki/Hard_disk_drive_failure - fair outline of what makes a drive die


EDIT:
BB's FAQ - https://www.backblaze.com/blog/hard-drive-stats-faq/

Q: Why do you hate Seagate?

A: We love Seagate! This is a common misconception about our hard drive stats posts. Both of our hard drive stat posts indicated a slightly elevated failure rate from their compatriots, however, the price of the Seagate drives cannot be beat, and they work great in our environment. This goes back to the price to density to failure-rate ratio that Backblaze likes to use, and in it, Seagate often comes out on top!

BB on enterprise drives - https://www.backblaze.com/blog/enterprise-drive-reliability/

 

[copenhagen69]

Perfect, thanks @claes for the abundance of info ... I got a lot of reading to do haha

 

[SuperSoph_WD]

Hey there, guys!

I think it's only fair to contribute with just a little more details about how drive reliability is measured in WD HDDs and help out @copenhagen69
Basically, we no longer measure the reliability of our hard drives using Mean Time Between Failure (MTBF). We use Component Design Life (CDL) and Annualized Failure Rate (AFR) instead.
And the results has shown that the Component Design Life of the drives is 5 years and the Annualized Failure Rate is less than 0.8%. You can see this in this article from our KB: http://products.wdc.com/support/kb.ashx?id=0LSeMS

However, you know that the mechanical hard drive is a very sensitive hardware, which is why doing regular backups is the best way to avoid significant data loss. Whereas the limited warranties give you some peace of mind that you'd be provided with a replacement soon enough. You can also take a look at our warranty policies worldwide here: http://products.wdc.com/support/kb.ashx?id=9ADmd1

Hope this was useful! Good luck!
SuperSoph_WD

 

[R-T-B]

HGST Ultrastars (A Western Digital Company now) are totally awesome. I'm not so fond of their old Deskstar drives, but I think Toshiba handles that now.

SuperSoph, is there any reason the simple MTTF (Mean Time To Failure) isn't used? MTBF has always been a bit fuzzy how they calculate it but I'd think simple MTTF would be easy to understand.

 

[PLAfiller]

People above me, owned way way more HDDs than me. My humble experience is, I've got a WD black 0.5 TB, running in one of the laptops. Approaching 2 years, I am pretty satisfied with it. The blue series though- not so much. I had one die on me, with not much of reason I think. I'd even go to say Blue and Green are not worth it generally. Especially green.

I had to buy a new 2.5 inch drive last month and this time around I went with 1 TB Hitachi (HGST) - as mentioned a subsidiary of WD anyway. It was pretty good price. The seller really wanted to sell me 1 TB WD Black. The main argument being 5 year warranty....but somehow I wasn't up for spending the premium over a regular 1 TB 2.5" drive. There's Seagate and Toshiba there too. I hope the Hitachi holds well.

 

[FordGT90Concept]

And the results has shown that the Component Design Life of the drives is 5 years and the Annualized Failure Rate is less than 0.8%.

That's sad. Most of my drives are over 10 years old. Designing for five years...

As to the OP, I trust nothing less than 7200 RPM Western Digital Black/RE or Seagate Barracuda/ES drives. Too many horror stories from the power saving drives.

 

[SuperSoph_WD]

HGST Ultrastars (A Western Digital Company now) are totally awesome. I'm not so fond of their old Deskstar drives, but I think Toshiba handles that now.

SuperSoph, is there any reason the simple MTTF (Mean Time To Failure) isn't used? MTBF has always been a bit fuzzy how they calculate it but I'd think simple MTTF would be easy to understand.

Hey, @R-T-B

Well, the reason is that we believe this is the best way to measure WD drives’ reliability.
However, the people who make those decisions at Western Digital are nowhere near me to ask, so unfortunately, I can't provide you with a more accurate answer at the moment. :/

SuperSoph_WD