In this case, though, the drive is CMR, so it shouldn't be horrible.So I'd almost treat this as a drive good for 'write once, read many' (WORM) type applications.
They might be fine with the speed, but these drives can't handle 24/7/365 usage. One of our clients put some Barracudas in their NAS and they were dead in under 6 months.most people would be fine with it.
Backblaze is slowly learning what the entire industry already knows--enterprise stuff is far more reliable than consumer. Over the various generations of their storage pods they've been moving more and more towards enterprise gear, first on the controllers and now on drives. I think their experiment is a bust in reality just showing that if you want real reliability it is with SAS and enterprise level drives and related hardware. In another decade I'm sure they'll come to the same conclusion, lol.I cannot speak about this specific model but in the past (8..12TB sizes) Barracuda closely matched reliability of Exos.
See for example stats for ST8000DM002 vs ST8000NM* in recent Backblaze report:
![]()
Backblaze Drive Stats for 2024
The 2024 Year End Drive Stats Report is here. See the the latest annualized failure rates and data insights for the Backblaze drive fleet.www.backblaze.com
Backblaze did switch lately to buy only Exos (when Seagate), though given their shucking history i suspect they're not paying much extra if at all for that (vs Barracuda).
I think choosing right vendor has more profound effect on reliability than difference between enterprise and desktop drive models (talking SATA/capacity drives here).
Myself - I no longer have any Seagate drives. They had a number of models with dramatically high failure rates (AFR 5-8-12%); (i've lost 4 of the famous ST3000DM001 within a year); HGST/WDC and even Toshiba do not come close, usually fluctuating under 1%.
If a drive is delivered with SAS or SATA interface does nothing towards its reliability or expected lifespan tho.And just the simple move to SAS just changes the game as those drives are tremendously reliable
Yeah pretty much all the tempting storage nodes coming out of scale deployments are SATA only now.SATA is just as fine as SAS, with the exception of many bad SATA cables. When Dropbox moved off AWS and to their own DC, they used SATA drives.
It shouldn't, but ime it does. And I've got such a cross section of different applications that it could only be the difference in drive quality.If a drive is delivered with SAS or SATA interface does nothing towards its reliability or expected lifespan tho.
The easy way to find out is to push 2x drives, one of each, to a duty cycle 2-3x its rated capability--the weak ones will fall out early and the lesser quality ones will show later.Beyond the CMR/SMR differences I really question how much is marketing and how much is actual technical differences between the "enterprise" and "consumer" lines of spinning rust. We already know they are (or used to) purposefully mislabel 7200 rpm drives as 5400 to have one fewer production line.
I was more talking about facts and reality tho, not your feelings that are not based in either of them.It shouldn't, but ime it does. And I've got such a cross section of different applications that it could only be the difference in drive quality.
That's facts for me. And you can take your feelings on the matter elsewhere if you want to have that attitude.I was more talking about facts and reality tho, not your feelings that are not based in either of them.
But its not like you are alone in this.
There are some facinating splits like this between enthusiast and "real world" enviroments.
Where what is acceptable to use in enterprise enviroments without any concern is beneath what some enthusiasts deem acceptable.
You would have been correct to a degree 10-15 years ago, then the sas and sata drives would be from seperate production lines.
But for quiet some years now its the same drives that are used for both, there is no seperation before the controller pcb goes on interface its sold with.
sas usage when the failure domain is at node/rack/row type scale has dropped like a brick.
The world has moved forward, but parts of the enthusiast community like you has not.
They are the same upto controller board with its interface/protocol is added, the sas one typically uses around 3w more.If the drives are so much the same, why is the power spec for sas different? If it's the same drive motor and hardware
Unfortunately this claim is useless. Barracudas and Barracudas can be drastically different.They might be fine with the speed, but these drives can't handle 24/7/365 usage. One of our clients put some Barracudas in their NAS and they were dead in under 6 months.