Backblaze Hard Drive Stats Q3 2020 (backblaze.com) 37
Backblaze's Q3 2020 hard drive stats: As of September 30, 2020, Backblaze had 153,727 spinning hard drives in our cloud storage ecosystem spread across four data centers. Of that number, there were 2,780 boot drives and 150,947 data drives. This review looks at the Q3 2020 and lifetime hard drive failure rates of the data drive models currently in operation in our data centers and provides a handful of insights and observations along the way. [...] There are several models with zero drive failures in the quarter. That's great, but when we dig in a little we get different stories for each of the drives.
The 18TB Seagate model (ST18000NM000J) has 300 drive days and they've been in service for about 12 days. There were no out of the box failures which is a good start, but that's all you can say.
The 16TB Seagate model (ST16000NM001G) has 5,428 drive days which is low, but they've been around for nearly 10 months on average. Still, I wouldn't try to draw any conclusions yet, but a quarter or two more like this and we might have something to say.
The 4TB Toshiba model (MD04ABA400V) has only 9,108 drive days, but they have been putting up zeros for seven quarters straight. That has to count for something.
The 14TB Seagate model (ST14000NM001G) has 21,120 drive days with 2,400 drives, but they have only been operational for less than one month. Next quarter will give us a better picture.
The 4TB HGST (model: HMS5C4040ALE640) has 274,923 drive days with no failures this quarter.
The 18TB Seagate model (ST18000NM000J) has 300 drive days and they've been in service for about 12 days. There were no out of the box failures which is a good start, but that's all you can say.
The 16TB Seagate model (ST16000NM001G) has 5,428 drive days which is low, but they've been around for nearly 10 months on average. Still, I wouldn't try to draw any conclusions yet, but a quarter or two more like this and we might have something to say.
The 4TB Toshiba model (MD04ABA400V) has only 9,108 drive days, but they have been putting up zeros for seven quarters straight. That has to count for something.
The 14TB Seagate model (ST14000NM001G) has 21,120 drive days with 2,400 drives, but they have only been operational for less than one month. Next quarter will give us a better picture.
The 4TB HGST (model: HMS5C4040ALE640) has 274,923 drive days with no failures this quarter.
the data is conclusive... (Score:2)
Re: the data is conclusive... (Score:2)
Re: (Score:3)
Re:the data is conclusive... (Score:5, Insightful)
make backups.
That is not what the data says.
The data says that HDDs are more reliable than ever.
Re: (Score:1)
Re: (Score:1)
The more and louder one repeats some statements without any arguments, the more one disolays that one does not actually know.
Re: (Score:2)
Are you arguing that people *shouldn't* make backups?
Re: (Score:3)
The data says that HDDs are more reliable than ever.
True... but the data indicates they are not perfect. Make backups.
Re: (Score:1)
Nothing is perfect.
There is a chance of 1 in 10^58 that all your atoms miss all the floor atoms and you just fall through the floor.
The chance of you dying today, is much, MUCH higher.
Re: (Score:2)
I don't think the data says that at all. It looks like it's just simply looking at whether the drives are basically a dud pulling out of the box. My understanding is that most of the drives have been only running for a few days, and you have to divide the drive days by the number of drives. There are plenty of drives listed in the article that have AFR of more than 1%.
Re: (Score:2)
The summary says that. Backblaze's data is actually far more detailed and includes far more than 5 models of drives. That data actually has shown that drives have gotten more reliable since Backblaze has started publishing meaningful data.
Re: (Score:2)
Re: (Score:2)
Still make back ups. Things can fail at any time. Even bad luck.
Re: (Score:2)
The data says HDDs still fail sometimes, so make backups.
Unless the failure rate is guaranteed to be zero, make backups. And even then don't trust it, make backups.
Re: (Score:3)
It certainly is Backblaze opinion.
Other than that, drive failures are more of a case for RAID than backups. Backups are important for more reasons than drive failure: accidental deletion, malware attacks, bugs, disasters, etc... but they are not ideal if you just consider drive failure. Restoring from backup has a cost: you may still lose some recent data, and your data is unavailable during the restoration process.
A proper RAID-type system is much better at dealing with drive failure. It is essentially tra
Question (Score:2)
This is only about spinning drives. Why not SSDs? Is there a reason these "slower" drives are used rather than SSDs? Do they have SSDs in their inventory but just don't report them?
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
the cost of my first pair of 80 Meg SCSI HDs
You and the other 999 customers from that time can have a party with the 1000 people that would spend $1800 on 18TB in 2020.
Re: (Score:1)
Re:Question (Score:5, Informative)
Re: (Score:2)
LTO tapes are cheaper still @ $7 / TB.
For backup tapes are perfect, you can move them to an other place, the do not need any power.
Re: (Score:2)
Re: (Score:1)
Uuum, 3-3-3-3 would make more sense.
3 sites, each running 3 as-different-as-possible systems with with triple-redundant drives, read and written by 3 separately developed, completely independent implementations.
Ideally, the higher the number, the higher the sigma of statistical reliability.
Re: (Score:2)
Re: (Score:2)
LTO tapes are cheaper still @ $7 / TB.
But don't support random access making them completely useless in any application other than for making backups.
"Cloud backup" Is not about duplicating data. It's about data redundancy which can be accessed randomly. Hell just the other day I accidentally nuked a file by overwiting it with the wrong data.
Option A: Get most recent LTO tape, go through incredibly long process finding the data on the tape, and recovering it, and hoping that the backup is less than 1 day old (it wouldn't have been, my offline b
Re: (Score:2)
It's also worth noting that their data is based on datacentre conditions, i.e. air conditioned racks full of actively cooled drives that are spinning 24/7.
A home NAS probably spins the drives down and back up several times a day. Cooling may be better, it may be worse (remember to clean your fans). Power in the data centre is probably a lot cleaner than in your home.
The spin up/down cycles in particular can be a cause of failure. It puts strain on the motor and causes the drive to thermally cycle between ho
WD (Score:1)
Are they avoiding Western Digital?
Re:WD (Score:4, Informative)
Are they avoiding Western Digital?
Only because it doesn't meet their criteria; nothing particular against WD itself. From a comment [disq.us] on their Q2 stats, where someone noticed the same:
We buy drives based on price and availability. We didn't stop buying WD drives, they just couldn't meet one or both of these criteria. We do buy HGST drives, which is owned by WD, but as on the end of Q2 we did have have any of the WD branded drives in operation.
Toshiba drives = HGST (Score:1)
They are HGST internally.
Which is the former IBM division by the way.
They made a 180 degree turn after their debacle with the last IBM-branded ones.
Yes, they are very reliable nowadays, as far as I can tell.
EPA should take actions against Backblaze (Score:2)
Re: (Score:2)
You're making assumptions that they didn't take this into consideration and mount every other drive upside down.
What? (Score:2)
The numbers don't make any sense.
If you have 150,XXX drives with 2XX,XXX drive-days, it seems you installed the drives the day before yesterday.
What am I missing??
That's amazing (Score:1)
"The 4TB HGST (model: HMS5C4040ALE640) has 274,923 drive days with no failures this quarter."
Does this mean they have 274,923 of these drives and not a single one died in 4 months?
I didn't think that degree of reliability was possible.
Re: (Score:2)
274,923 drive days ~= 2290 drives for 120 days
And of those 2290, they had no failures. Still impressive, imho.
Re: (Score:2)