Have Hard Drives Become Obsolete???

[josephkhan]

Let me start off with a little story:

I have a seagate external 5 TB drive I bought for a steal from costco a while back. It stopped responding in both Linux and Windows. Rather than dealing with the warranty process, I decided to shuck it and install it in my desktop (it's the only spinning disk in my system, everything else is solid state).

It worked. I could see the drive. All the files are gone, but I don't keep anything important stored on drives like this anyway. I reformatted it. Then later decided to check it for bad sectors using chkdsk in Windows. It took 10 hours! Yes it found bad sectors. 3520kb worth.

That's insane. Every drive eventually develops defects. I've had bit rot and occasional bad sectors occur on drives from every brand. A note to all of you who are new to hardware, NEVER store important data in only one location, especially a hard drive. Especially if you use compression tools like WinRAR or 7-Zip or disk imaging tools like Macrium Reflect. If a bad sector develops or bit rot occurs (and it will, without a doubt) you WILL lose data.

Which brings me to my point. I've yet to have an SSD die. Even the earliest sandforce based ones I have are still running strong. No bit-rot either. Have we reached the end of usefulness when it comes to spinning disk based storage? Sure they are priced lower, but they are so unreliable that at best they should be considered 'temporary' storage.

I moved to an SSD only system some time ago, but I keep a few external drives around to mirror my cloud backup solution or store media on (ripped discs, etc.) However, SSD prices are projected to continue to fall, so while SSDs may never match hard drives in price, they are affordable for most people, and they are much more reliable.


Note: This post was plagiarized word-for-word from Cnet

[NookieN]

Yes, you should always store important data in more than one location. This also includes SSDs. I have had a SSD fail, and I have coworkers who have had SSDs fail. It's very rare, and as you say they are far less prone to sector corruption. In a large enough sample you will have some fail.

[nomead]

It all depends on a number of things. How much data you have to store is one thing. Even "cheap" SSDs aren't that cheap for multiple terabytes yet. And I've had one SSD fail thus far. A Samsung 830 that was about 4,5 years old at that point so in principle still within warranty, but meh. So small that it wasn't worth getting a replacement through the warranty program, I just bought a bigger 850 PRO instead. Luckily it wasn't catastrophic, the data was still readable but any writes were either very slow or didn't go through at all.

One thing that SSDs don't handle well at all is "cold" storage. Unpowered data retention is in the range of months, max. a year, if you can even find it in the specifications nowadays. While powered it isn't a problem. The drive constantly reads back data on the drive and rewrites blocks that are marginal, i.e. bits are flipped before error correction. And this is getting ever more important with TLC and now QLC flash memory. The error correction algorithms running in the background are really something...

[CRGreathouse]

SSDs fail, spectacularly. One problem is that deleting data is very difficult, and requires the application of large voltages. This is intrinsically damaging. For consumer SSDs, you can notice a diminution of capacity as damage appears; for 'professional' drives, you won't necessarily notice because they're typically overprovisioned to account for this.

Edit: but hard drive performance is bad, so I imagine everyone will move away from them for all purposes except cold storage of data and maybe a few other specialized purposes.

[aurashift]

Not yet. Not completely. They are losing ground though.

[Mysticial]

I've had 3 SSDs die out of about ~25.

The first was an OCZ Vertex - yes those...
The 2nd and 3rd came in the past two years at work. They both had Intel SSDs subject to normal desktop workloads.

The OCZ failure was expected since they were dying for everybody. The two failures at work were probably just bad luck.

Two of the failures (OCZ + an Intel), died catastrophically after acting up once or twice. (i.e. drive not detected)
The other Intel never completely died, but acted up badly enough that the system was unusable as it would constantly hang on stalled requests to the SSDs.

-----

By comparison, my HD failure rate is about 8 drives out of ~50. This also seems high but were biased high due to overall age and from one particular Seagate model that Backblaze reported double-digit failure rates. (I had 5 of them, 4 of them died within 2 years.)

[ewmayer]

Much also depends on the physical context and data volumes involved - mobile devices, even the 'shock-proof' ones, you absolutely want no moving parts. Huge-volume storage, OTOH, a spinning-disc RAID is still the most cost-effective solution. Of course there the 'R' is key.

But I expect within 10 years spinning-discs will have become a fading niche item, rather like photographic film is today.

[kriesel]

Originally Posted by ewmayer

Much also depends on the physical context and data volumes involved - mobile devices, even the 'shock-proof' ones, you absolutely want no moving parts.

I disagree. The manufacturers long ago started including park-on-fall-detection. This 10 year old laptop took enough 3-foot dives to destroy the clamshell hinges, but the original HD is still going strong. iTunes players and some digital cameras employed physically small hard drives, that fit in a compactflash form factor or smaller. (I think the record for HD miniaturization was something like 0.7 inch platter diameter. Toward the large extreme, I have a HD platter hanging in my hallway that measures 39." diameter.)

[retina]

There is nothing wrong with HDDs. Their reliability is perfectly fine. Their robustness is perfectly fine. Their speed is perfectly fine. Their capacity is far superior.

SSDs have no advantage over HDDs except for speed. That is all. They are not more reliable, or more robust. Their capacity is awful.

I have banned SSDs from my lairs systems. Too small. Too expensive. And not enough faster to make any meaningful difference. I don't care if a machine boots 10 seconds faster, it's going to be running for months between restarts so I'd much rather have the data capacity (and more money in my pocket) than save 40 seconds each year.

[ewmayer]

Originally Posted by kriesel

I disagree. The manufacturers long ago started including park-on-fall-detection.

That doesn't guard against a fall or other shock from physically damaging some part of the movable r/w system inside the HD. My last macbook HD failed catastrophically from a modest fall-to-hardwood-floor event, not its first, which fact had lulled me into a false sense of security. After the aforementioned terminal event, one could hear a repetitive tick-ticking noise, but without any signs of data readage. The replacement drive is another spinning HD because that's what I happened to have on hand (with desired version of OS installed but lacking any personal data), but since then I do regular backups to a bootable mirror-clone SSD. If/when the HD fails in will go the SSD, and I will use one remaining HD as the backup for that. In fact I may do the swap sooner that that because waiting ~60 seconds for the HD-based system to come out of hibernate mode, multiple times per day, gets quite wearisome.

It'll be fun to revisit this thread in 10 years and see where things lie for the 2 respective technologies.

[kriesel]

Originally Posted by ewmayer

That doesn't guard against a fall or other shock from physically damaging some part of the movable r/w system inside the HD. My last macbook HD failed catastrophically from a modest fall-to-hardwood-floor event.

Well there's your problem; my laptop has been diving to carpet. And I've always had an 18 hour memory gap from when my head hit the asphalt on a bike ride; sod or sand or a pillow hasn't been a problem. The tick-tick-tick is I think the can't-seek sound. Parking the heads gives the data recovery pros a better shot, since the head crash if it occurs is probably where your data isn't, so it's not scraping your ones and zeros off the platter and shuffling them in the case. For a price, they'll open it in a clean room and another sacrificial drive like it and graft the precious data-filled platters onto a working read/write mechanism and electronics.
I think you're right that HD usage will diminish; solid state has taken over in small size applications like cameras, mp3 players, etc. and will expand its turf as capacities grow.