• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

How soon until magnetic HDs become obsolete?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Cerb said:
Depends on temperature and wear. IMFT offers MLC rated by them for up to 5 years at ~70F (or, at least, they did though 25nm, and projected the same for the future). JEDEC spec is 1 year. Those times are based on being at the p/e rating.

So, yes, but the temperature sensitivity is what mainly sets them apart from HDDs. Intel is the only one I know of that's shown research data from testing at different temps, and at temps that a drive might find itself in your car on a hot day, they were down to just weeks.

I was also quite ignorant of that not long ago, as there seemed to be somewhat of an urban mythology surrounding it for USB thumb drives.
Click to expand...

I'm running into this problem myself, it seems.

I have a pair (well, one is for a friend, and is installed, but in the box) of Foxconn AT-5570 AMD C-70 NanoPC units. They are book-sized PCs, which have a multitude of I/O connections (more than a NUC or Brix), and they also have both an mSATA slot, as well as a 2.5" drive bay. They are passively cooled. They stand up, and have an aluminum chassis, with vent holes on the top and bottom. The CPU is located at the top, with a heatsink. The 2.5" HDD is in the middle, towards the bottom.

Anyways, I bought a pair of 240GB OCZ Vertex Plus R2 SSDs, refurb, from Newegg.

I installed one of them into the AT-5570. After a few weeks, I was seeing speckles of red sectors in a HDTune surface scan graph. Eventually, Windows froze up, and the SSD had to be cleaned. This happened twice.

So I finally managed to secure-erase the 240GB refurb SSD, and hooked it up to an AMD Thuban 1035T rig, with plenty of cooling, and doing distributed-computing (thus insuring a steady stream of writes to the SSD). I had it running in that rig for a month. No red sectors!

But I took a 30GB OCZ Agility SSD, which had been used down to 75% health (10TB written), which was purchased new and used on several of my rigs for more than a year without issue, and put that into the AT-5570. I also think it used a larger-featuresize NAND, being an older and smaller drive. Well, come today, I did a surface-scan using HDTune, and it has developed a red (bad) sector block.

So it seems that my theory that the high temps were causing data errors on my SSDs in these PCs, is true!

The CPU temp from CoreTemp shows a max of 77C. I remember last time that the drive failed, and I removed it, the SSD was scorching hot, it seemed. I wouldn't be surprised, that with the passive cooling, and the proximity to the CPU, that the SSD might get to 60-70C. Just a thought. I can't get temp measurements, because this rig runs AMD AHCI drivers.
 
Cerb said:
Depends on temperature and wear. IMFT offers MLC rated by them for up to 5 years at ~70F (or, at least, they did though 25nm, and projected the same for the future). JEDEC spec is 1 year. Those times are based on being at the p/e rating. So, yes, but the temperature sensitivity is what mainly sets them apart from HDDs. Intel is the only one I know of that's shown research data from testing at different temps, and at temps that a drive might find itself in your car on a hot day, they were down to just weeks. I was also quite ignorant of that not long ago, as there seemed to be somewhat of an urban mythology surrounding it for USB thumb drives.
Click to expand...

not going to work with the computerization of everything

they need to rate those at least to 130F
 
norseamd said:
not going to work with the computerization of everything
Click to expand...
It has nothing to do with the computerization of anything, but specifically NAND intended for the consumer electronics markets. Other markets have different requirements, and products to suit them exist, though not at the same low prices.
they need to rate those at least to 130F
Click to expand...
Why? If you leave your SSD in a non-climate-controlled storage unit, or car, or desert hut, for several months, having hammered it to anywhere close to its rated endurance, and expect everything to be peachy, the problem is not the SSD. If stored in a climate-controlled environment, all will be fine. It's probably not the best treatment for a well-used HDD, either.
 
This is like asking, "More and more people are using mobile phones and tablets! How soon before keyboard-and-mouse computing becomes obsolete?" Or, "More and more people are using wireless networking! When will wired networking go the way of the dinosaur?"

To better highlight the inanity of the question, let's try, "I see more and more casual Facebook-style games. When will they conquer the entire gaming market and render those games with pesky GPU requirements obsolete?"

Each has its strength and weaknesses and are suitable for different things. There are areas of overlap, and there you can see one product being displaced by another. But it's an exercise of faulty extrapolation to extend that across the board.

There will always be hard drives. They're far easier and cheaper to make for the capacity they provide. Yes, we're hitting a wall with respect to magnetic density, but that's true for semiconductors, too--and as cell sizes shrink, there will be more negative pressure on things like longevity and data retention. There is little reason to expect SSDs to ever be competitive with HDDs in terms of price-per-bit. I do expect SSDs to take over "primary" storage--OS, apps, etc. But not extended storage.

(I have 5 computers all with SSD boot drives, and I even upgraded the boot drives of my relatives' computers to SSDs. Yet, just months ago, I bought 16TB of HDD storage. Each has their own domain.)
 
Last edited:
Just my lightroom catalog is right around 500GB. It's simply not practical to use SSD's for backup. By the time they have affordable 1TB drives then I will have a much larger catalog.

I don't need a SSD to backup TB of data.

$550 for 1TB of SSD or practically nothing for a mechanical drive. I wonder which one 99.99% of people will chose for the storage needs.
 
Cerb said:
Why? If you leave your SSD in a non-climate-controlled storage unit, or car, or desert hut, for several months, having hammered it to anywhere close to its rated endurance, and expect everything to be peachy, the problem is not the SSD. If stored in a climate-controlled environment, all will be fine. It's probably not the best treatment for a well-used HDD, either.
Click to expand...

then what are you going to use for storage for electronics in vehicles in the tropical areas?
 
I'll say it again . . . HDDs are obsolescent but not yet obsolete. There is a difference in those terms. The question is, when will they become obsolete?
 
norseamd said:
what about vibration prone vehicles like off road buggies
Click to expand...
Off-road vehicles will have something aftermarket, which will be whatever you want, or whatever some device you use supports. But, again, why is it the storage devices fault that you leave it out there in such environments, powered off, and then expect it to be OK? It's not the SATA SSD's fault, or SD card's fault, or USB thumb drive's fault, if you leave offline backups in a vehicle exposed to the elements.
 
silicon said:
+1 to that. SSD's will improve, become cheaper and will leave HD's i n the dust.
Click to expand...

That's blind faith in the extrapolation of Moore's Law. As I said, semiconductor shrinkage is getting close to hitting its limit. So, too, is magnetic shrinkage. A simple matter of physics. Frankly, I don't think see how it is physically possible for solid state storage, where every bit must have a physically-manufactured structure, to ever be as cost-effective to produce than magnetic storage.

I feel like a broken record here: There are places where SSDs will prevail. Situations where speed is important (boot drives) or where physical durability is important (anything with lots of movement and acceleration). That doesn't mean that, by means of magic fairy dust, SSDs will also conquer areas where cost-per-bit is the most important thing. I have SSDs in every system, yet I sure as hell am not going to use SSDs for my 16TB file server.


As for off-road vehicles, keep in mind that NASA spacecraft use solid-state storage. Now how's that for extreme? With the proper hardening and other measures (using SLC, a larger manufacturing process, generous amounts of Reed-Solomon redundancy, etc.), it can survive those tougher situations. Just because consumer-grade solid state storage that's more focused on cost (e.g., MLC/TLC, tiny NAND cells) isn't as durable doesn't mean that we can't make something more durable (though more expensive) for certain niche applications.
 
Last edited:
code65536 said:
There will always be hard drives. They're far easier and cheaper to make for the capacity they provide. Yes, we're hitting a wall with respect to magnetic density, but that's true for semiconductors, too--and as cell sizes shrink, there will be more negative pressure on things like longevity and data retention. There is little reason to expect SSDs to ever be competitive with HDDs in terms of price-per-bit. I do expect SSDs to take over "primary" storage--OS, apps, etc. But not extended storage.
Click to expand...

HDDs are cheaper now, but they have basically hit their scaling limit a few years ago, and since have scaled much, much slower.

NAND flash based memory on the other hand is still scaling rather well. The ceiling on these is at least another 5 years away, and there are still breakthroughs such as triple level cells being added, which further increase storage per area. Furthermore, 450mm fabs should in theory make flash memory cheaper yet, and the lack of maturity of the domain makes me hopeful for some non-process-based technology to still be accessible.

Once SSDs hit the price point of HDDs per GB, there will be very little reason to still buy magnetic disk media. Already in the budget sector, this scenario isn't so far out - the cheapest SSDs are approaching the cheapest HDDs in price/memory, at least in the retail market.

On the other hand, what might obsolete HDDs at around the same point in time is an actual gigabit Internet connection at home. That would allow massive deduplication, and would reduce the actual need for storage, while requiring central storage with very high read rates.

I would put either around a decade away (which should get us to well into single digit nm resolution flash cells, and allows for FTTH to become much more common)
 
It's going to be quite a ways into the future before HDD's completely disappear. SSD's will have to reach price parity and they'll have to be able to retain the contents of memory for longer time periods (not that it's bad now) without being powered on. The more important thing is continuing the trend toward using SSD's for booting and programs. HDD's are more than adequate and make more sense just for mass data storage.
 
The one thing I've wondered about SSDs with regards to capacity, is what's stopping manufacturers from creating SSDs in 3.5in or even 5.25in form factors, and actually utilizing that extra space for more memory cells?
 
_Rick_ said:
Already in the budget sector, this scenario isn't so far out - the cheapest SSDs are approaching the cheapest HDDs in price/memory, at least in the retail market.
Click to expand...

The cheapest per-GB SSD is around $0.50/GB. The cheapest per-GB HDD is around $0.03/GB. We're still off by over an order of magnitude. Shingled recording will add a bit more scaling to HDDs, and as for SSDs, we're already at 19/20nm, and the limit of scaling is... what... 7nm or thereabouts? But more importantly, we're almost at the point where scaling doesn't result in cost reductions per transistor because of the sheer difficulty of the process (e.g., each wafer requiring many more steps).

Oh, and back in late 2012, I was buying SSDs at $0.50/GB (got two HyperX 3K's at that price) and even grabbed a low-end model for $0.42/GB. Over a year later, the prices are about the same--they only seem cheaper because SSD prices actually increased in the interim.
 
Golgatha said:
The one thing I've wondered about SSDs with regards to capacity, is what's stopping manufacturers from creating SSDs in 3.5in or even 5.25in form factors, and actually utilizing that extra space for more memory cells?
Click to expand...

Cost. How many dies you clump together into a single drive unit doesn't magically make those dies cheaper to manufacture. If they really wanted to, they can pack enough dies into a 2.5" drive for 4TB. The reason they don't offer that is because they figure (correctly, I might add) that the market for a $2000 drive isn't exactly very big.
 
No doubt that mechanical hard drives will be obsolete at some point, I think the gist of the original question is when?
When I first started working on computers, the computer's storage drive was the size of a small car wheel and probably held about 10MB of data. Now, I have SSD's in all my computers. That's about forty plus years of technology. I'm going to say it's going to take at least another ten years for current hard drive technology to be discarded for something better. What that is, I don't know. Whatever it is, it better be cheap because I have a lot of hard drives with a lot of data I need to store.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top