Backblaze shares 2017 HDD reliability stats

[HEXUS]

Cloud storage data centre company relies upon over 90,000 HDDs.

Read more.

[gagaga]

I think the short conclusion is wow, consumer drives are incredibly reliable now.

From my own server, i've had 15 8TB seagate drives running for over three years without a single failure - that's 16,000 drive days. Maybe I should publish my own stats?

[DaMoot]

Consumer drives have been reliable for a long time. In 20 years, and dozens of hard drives, I've only ever had 2 failures. Only one of those was a catastrophic failure.

I've seen more failures at work in the past 12 years with my current company with so-called Enterprise-grade drives than I've ever seen in my personal and professional experience with consumer drives.

I think the takeaway is that yes, drives are extremely reliable for easily the past 15 years, and WD still isn't sure it knows how to make a reliable hard drive! lol

[Ozaron]

My last WD Blue 1TB drive failed after barely more than a year (of light use) while having a virus scan run on it. Maxed out at around 260GB stored. Welp.

[chinf]

I have a small population of 3.5" hard drives of varying ages, including Seagates, Samsungs, HGST and WD, and the worst performers have been the Seagates by a country mile.

Among my most recent failed drives were two Seagate ST2000DM001s which saw little use and barely lasted a year each before failing with very quick degradation, but my six ancient 160GB IDE Barracudas are still working fine though not used much anymore. I've had a 1TB WD Green slowly fail that was in constant use for several years and lasted well beyond warranty, and I've got several other WDs of varying ages and they're fine. No problems yet with my four 500GB Samsungs even though they're now pretty old.

Not sure if Seagate will fix their reliability before spinning rust becomes extinct.

[shaithis]

Not sure if Seagate will fix their reliability before spinning rust becomes extinct.

I doubt it, the STxxxxDM001 drives have been around for a number of years now and have been shoddy since day 1, can't see that changing now.

[DanceswithUnix]

I think the short conclusion is wow, consumer drives are incredibly reliable now.

From my own server, i've had 15 8TB seagate drives running for over three years without a single failure - that's 16,000 drive days. Maybe I should publish my own stats?

They are supposed to be reliable as long as they are in warranty.

... with so-called Enterprise-grade drives than I've ever seen in my personal and professional experience with consumer drives.

Whose enterprise drives are those then?

I think I have seen failures from every drive manufacturer past and present. Remember Quantum? They were really good, but I still had a consumer one where the head fell off.

[CAT-THE-FIFTH]

They are supposed to be reliable as long as they are in warranty.



Whose enterprise drives are those then?

I think I have seen failures from every drive manufacturer past and present. Remember Quantum? They were really good, but I still had a consumer one where the head fell off.

Quantum Fireball!! I trust Seagate the least of all the hard drive brands so far TBH.

[DanceswithUnix]

Quantum Fireball!! I trust Seagate the least of all the hard drive brands so far TBH.

My first ever hard drive was a Seagate SCSI 30MB (not a typo, megabyte). It was a long time ago, but I think it went back under warranty three times. Some people get nothing but joy from Seagate, but my experience with them hasn't improved.

[CAT-THE-FIFTH]

My first ever hard drive was a Seagate SCSI 30MB (not a typo, megabyte). It was a long time ago, but I think it went back under warranty three times. Some people get nothing but joy from Seagate, but my experience with them hasn't improved.

Samsung used to be awesome until in their later years,the reliability just went south. I think the oldest drive that still is probably working is an old 7200RPM IDE WD drive,which ran very hot but kept on going and going.

[gagaga]

I think the short conclusion is wow, consumer drives are incredibly reliable now.

From my own server, i've had 15 8TB seagate drives running for over three years without a single failure - that's 16,000 drive days. Maybe I should publish my own stats?

They are supposed to be reliable as long as they are in warranty.

Warranty is 12 months (at least on the early ones) so they are way out of it.

Of maybe 80 drives i've owned over the years i've only ever had 4 fail - an Hitachi 3TB (ironically of the type Backblaze have found great), a WD 2TB Green and two laptop drives (a Samsung that started losing data and an IBM(!) that sat below the thin plastic wrist rest on a laptop and got squashed.

My server used to live in the loft and got *hot* in summer (the Hitachi died during this time) .. it now lives in a draughty, damp, unheated cellar and gets cold. Had no problems at all in the past 3 years it's been there with the drives spinning up just fine in sub-zero weather.

[Millennium]

I...

Among my most recent failed drives were two Seagate ST2000DM001s ...
...

Damn, I'm running 2 ST2000DL003 in Raid 0. Playing with fire! They've been ok on and off for a few years though Maybe SMART will alert me?

Slightly off topic (maybe) but I've recently got a portable (2.5") 4tb WD My Passport drive for backups mainly. I've heard that 2.5 is much less reliable though, is this true? Should I sell it off and get a full size one? Physical space isn't a problem.

[gagaga]

Damn, I'm running 2 ST2000DL003 in Raid 0. Playing with fire! They've been ok on and off for a few years though Maybe SMART will alert me?

Slightly off topic (maybe) but I've recently got a portable (2.5") 4tb WD My Passport drive for backups mainly. I've heard that 2.5 is much less reliable though, is this true? Should I sell it off and get a full size one? Physical space isn't a problem.

You're asking for pain with RAID0 or RAID5/6 in anything that is not dedicated and left permanently powered up ... I'd split asap unless you've got (very) regular tested backups.

2.5 - these are way more reliable in my experience for external drives. Especially for backup. If you knock a 2.5 drive whilst it's in use, or drop it you've got a very good chance (as in near 100%) chance it'll be okay as they have movement sensors and fast parking as a result of their design for laptops. Knock a working 3.5 drive and you've a good chance a head crash will spoil your day (they are stupidly robust when parked/spun down, however). I advise everyone to stay away from 3.5" USB drives. There's just no need to take the risk given you can get 5TB 2.5" drives on Amazon for £120 now (I paid that for a 30GB drive VAT free in Japan in 2002...).

[aidanjt]

You're asking for pain with RAID0 or RAID5/6 in anything that is not dedicated and left permanently powered up ... I'd split asap unless you've got (very) regular tested backups.

No RAID level, JBOD, or single disk arrangement is a safe replacement for backups. Either it's disposable data, or it should be backed up, preferably a local copy and a remote copy, period.

[Iota]

No RAID level, JBOD, or single disk arrangement is a safe replacement for backups. Either it's disposable data, or it should be backed up, preferably a local copy and a remote copy, period.

Raid 10 isn't terrible if you can afford that many drives, although nothing quite beats having physical backups kept on a regular basis for anything important. I must admit, in all of the years I've run Raid 0, I've only once fallen foul of that option (old Maxtor drives). Samsung drives have been fine, and I do still miss my old raptor drives which for their time were brilliant.

Roll on affordable SSDs at the capacities we now require.

[peterb]

Raid 10 isn't terrible if you can afford that many drives, although nothing quite beats having physical backups kept on a regular basis for anything important. I must admit, in all of the years I've run Raid 0, I've only once fallen foul of that option (old Maxtor drives). Samsung drives have been fine, and I do still miss my old raptor drives which for their time were brilliant.

Roll on affordable SSDs at the capacities we now require.

RAID is about resilience and maintaining uptime allowing a failing drive to be swapped out with minimal service interruption (the exception is RAID0 which just doubles the risk of data loss as failure of one drive can affect the data on both). They are not backup substitutes.

SSD don’t alter that - in fact they increase the need for backup because when they fail, they are (at present) more likely to fail catastrophically rather than giving early warning signs.