Who is knowledgeable about the nature of HDD failure?

[BonzaiDuck]

Probably bought my first 10MB WD HDD around 1988 or '89. It is an electro-mechanical device. The nature of the failure rate for SSD and HDD are quite different. There are no moving parts in an SSD. But you only get so many TBW before an SSD becomes a ROM.

I have this media drive in a the 2700K workhorse system of my sig. I use it to store DVR captures and provide a buffer for Media Center Live TV. Truth is, I moved "that function" from a 2600K after 5.5 years of solid duty, and I copied the hard disk, which is still part of the 2600K configuration. Saving DVR and movies that would play on the 2700K -- unencrypted content -- I added some new captures, including a whole week and five seasons of "Breaking Bad." There may have been a video capture that was being interrupted because I'd rebooted the computer -- I can't say.

The other day, I noticed my TV was stuttering and freezing, and I thought it was my Silly Dust tuners, my network, the scanned channel lineup -- any of those, and I checked them out. Then I realized that the HDD light specifically for the hot-swap media HDD was going super-nova: the drive was working hard. For? The Media Center Buffer of a couple GB.

Identified the drive as a WD 1TB Blue. Another drive in the box is a Samsung Spinpoint F1 (?) 1TB -- doing great things. Ran WD Diagnostics and the drive failed the Quick Test without too much lapse of time.

So far, I've almost got everything copied to safe storage except the Breaking Bad episodes, and I might as well buy the BD of the series, but I will try . . . .

How many times can you fill 60% of an HDD's space? How many re-formats and how much activity would it be good for? I'm sure the mechanical nature of the drive increases a probability of failure with the number of moving parts. But is there any limit to "TBW" for an HDD?

 

[Shmee]

Is it under warranty? If so, backup and RMA. If not, backup and toss.

 

[BonzaiDuck]

Is it under warranty? If so, backup and RMA. If not, backup and toss.

I'm trying to remember. this is 2017. I'm quite sure that I had these 1TB Blue drives as early as 2012. I'd used them mostly as backup, as media drives -- wherever I needed an HDD to supplement a system-boot SSD. I even have a couple of unused 500's and 320's. Blue drives could not have had more than a 3-year warranty. I was already comfortable with tossing the drive, although I think I'd best do some sort of secure erase before I take it to the recyclers.

Even I'm paranoid after the Traitor Snowden handed the secrets of NSA to Vlad Putin. Shred everything! Erase everything! People are rummaging through the city dump looking for your bank and credit-card statements!

I think the existing inventory includes about 3 1TB WD Blues still good, 2 Sammy Spinpoint F1's, four or five used WD Black 500's. I'm not at a loss for drive replacement. I just want to feel sure that my usage of the replacement disk or its age isn't going to cause me another 24 hour delay for copying files that I don't want lost.

Right now, I've got just about everything rescued except the Breaking Bad episodes. That could be another 24 hours of file rescue!

Afterthought: I also have a pair of unused Seagate NAS drives -- 2TB SATA-III's. They're for my server re-build -- in progress. I can get by with one of them until I complete that project and move the other four from the old server.

Also, I'm concerned these days, per media collections and the sizes of files, about having to back these things up. My network is Gigabit, but the old server is running off two Marvell controllers in PCI-E 1.0 slots with an 840 EVO system disk. I hadn't been confronted with "slowness" in other usage on that server, but I need to replace that old hardware sooner than later.

 

[VirtualLarry]

Most HDDs die from power-on time, and head retract / load cycles. Not TBW.

 

[lxskllr]

Most HDDs die from power-on time, and head retract / load cycles. Not TBW.

I lean more towards load cycles, specifically changing the drive's state. Almost everything does better when it does whatever it does continuously. Imo, running hot and hard is better than frequent powerups, with the associated surge, torque, and temperature changes.

 

[BonzaiDuck]

I lean more towards load cycles, specifically changing the drive's state. Almost everything does better when it does whatever it does continuously. Imo, running hot and hard is better than frequent powerups, with the associated surge, torque, and temperature changes.

I sleep and hibernate these systems to conserve electricity consumption. The only units that don't sleep include my workstation that happens to be functioning at that time as HDTV with WMC, and the server. the system with the drive going south currently hasn't slept or hibernated since converted to that purpose about three months ago.

The last drive that died was a VelociRaptor 650GB. or was it 600? Skinny passed to me by a retired e-tech suggests that particular VR capacity drive had tendencies to fail. I think it ran four years non-stop, and I got them to replace it under warranty.

I should do some defragging after this rescue mission is accomplished, and clone the drive, to the second Samsung F1.

Or maybe I'll use the 2TB Seagate spare I bought three months ago, still in sealed anti-stat wrap. It's a 2.5" drive that would fit in a laptop, but people use them for PS4 or Xbox -- something like that, as I recall reading before I bought the drives.

 

[grimpr]

I lean more towards load cycles, specifically changing the drive's state. Almost everything does better when it does whatever it does continuously. Imo, running hot and hard is better than frequent powerups, with the associated surge, torque, and temperature changes.

These are exactly my findings also, all of my drives that had run continously without power downs and load cycles did the best with trouble free operation.

 

[VirtualLarry]

I remember when WD Caviar (this is before their "rainbow of product lines") HDDs used ball-bearings, rather than the modern Fluid Dynamic Bearings. You could approximate the age of the HDD, by the loudness of the ball-bearing whine when powered-up and running, roughly to the year.

 

[rchunter]

These are exactly my findings also, all of my drives that had run continously without power downs and load cycles did the best with trouble free operation.

Me too, all my drives are running 24/7. Except for I keep a hot spare in each server all pre-checked and ready to go, the hot spares I keep spun down.

 

[Elixer]

There are no moving parts in an SSD. But you only get so many TBW before an SSD becomes a ROM.

I have never seen this happen, they all fail a miserable death.
The TechReport SSD life test clearly shows they all die and get tons of data errors, none of them turn into read-only devices.

How many times can you fill 60% of an HDD's space? How many re-formats and how much activity would it be good for? I'm sure the mechanical nature of the drive increases a probability of failure with the number of moving parts. But is there any limit to "TBW" for an HDD?

Not really, I agree with the others, head parking & power cycling cause the most wear & tear on HDs, assuming of course the temp is controlled, and there are no vibrations that introduce extra wear on the HD's parts.

 

[WhoBeDaPlaya]

How many times can you fill 60% of an HDD's space? How many re-formats and how much activity would it be good for? I'm sure the mechanical nature of the drive increases a probability of failure with the number of moving parts. But is there any limit to "TBW" for an HDD?

No limit. Any new / new-to-me HDDs get put through the wringer - at least 2 full rounds each of read and zero-fill tests, and it must do so without a single SMART error.

HDDs to be repurposed (eg. retired from file server to end-user system) gets a round of read + zerofill + SMART stat check.

Regardless of age / power-on time, as long as a drive has even 1 reallocated sector or pending reallocation, it gets junked / sold.

I have never seen this happen, they all fail a miserable death.

Or the translation table gets fux0red and the drive bricks itself.
Strangely, I've only had that happen with two supposedly bulletproof Intel X25M 80GB SSDs (and none of the dozen or so OCZs - go figure).