To all those that prophesied the demise of HDDs: That ain't happening anytime soon, son !

That we know. HDD are not going away, for a longer while.

Currently. The Seagate Ironwolf is such a god bit of kit. I saw an 8TB go for 100+- on ebay (new) the other day. I would have jumped in and got it but I didn't need it. 🙄 Long live the HDD. 🙂

wavetrex:

To all those that prophesied the demise of HDDs: That ain't happening anytime soon, son !

Not until high capacity SSD prices drop. Any SSD beyond 4TB is insanely expensive. Once they do though it's time for 10Gbit networking and a NAS upgrade though 😀

Spinning rust still has a place for now. can't argue with that. Solid-state tech is evolving much faster then magnetic. Solid-state may or may not be the tech to end magnetic permanently, but eventually the size and transfer speeds of magnetic will make it too risky a storage medium. My guess is that HDD will simply not be viable beyond certain capacities without major transfer speed improvements (beyond inherent density speed improvements).

Corrupt^:

Not until high capacity SSD prices drop.

The thing is, HAMR / MAMR are a big technology breakthrough, like perpendicular recording was (PMR), or in case of SSDs, like 3D cell stacking tech boosted sizes dramatically. I won't be surprised if HDD capacities will advance rapidly towards 100 TB, as well as read/write speeds, hitting SATA600 limits very soon, as HAMR increases bit density, resulting in more bits being accessed for the same rotation speed. I don't really see SSD catching up to that... actually, I don't see them advancing at all without cost increases, as they are kinda' reaching the minimum cell size that is still viable to store information reliable. NAND doesn't scale like CPU logic circuits... They can only increase the stack size, and even that gets harder and harder.

wavetrex:

The thing is, HAMR / MAMR are a big technology breakthrough, like perpendicular recording was (PMR), or in case of SSDs, like 3D cell stacking tech boosted sizes dramatically. I won't be surprised if HDD capacities will advance rapidly towards 100 TB, as well as read/write speeds, hitting SATA600 limits very soon, as HAMR increases bit density, resulting in more bits being accessed for the same rotation speed. I don't really see SSD catching up to that... actually, I don't see them advancing at all without cost increases, as they are kinda' reaching the minimum cell size that is still viable to store information reliable. NAND doesn't scale like CPU logic circuits... They can only increase the stack size, and even that gets harder and harder.

Doubt we'll see more then 500MB/s from HAMR 50TB. Also speed is further impacted by how far the heads must travel on the platter. Seeking at the outer portion of a platter is much faster then seeking to the inner. Even without DRAM cache, modern NAND writes are still far faster then what we'll see from magnetic any time in the near'ish future. That being said, you aren't wrong that it will get more difficult as time progresses to create more layers. However, a key difference between magnetic and solid state is that there is much higher need for more compact, energy efficient, and performant storage solutions. This means greater investment and competition with those solutions. NAND growth will gradually slow down and has done so a bit already but it's not going to hit the snail's pace that magnetic has anytime soon. We are in the race for more layers and better controllers now.

wavetrex:

To all those that prophesied the demise of HDDs: That ain't happening anytime soon, son !

I agree but drives like this ain't the reason. Unless they do something to improve sustained read/write speeds (even sequential), drives like this are going to be too impractical due to taking too long in the event of restoration from disk failure. Being mechanical (and a Seagate...), the probability of drive failure is high enough that it might make sense to just buy a smaller drive, possibly even an SSD. While there are use-cases for these drives, they're rather limited. Seems to me they're best used for archives. As for NAND, because it's just a chip on a PCB, it can take many form factors. So even if the data density isn't as great, there's already 100TB drives in a 3.5 inch form factor, from ExaDrive. The problem is of course pricing. So - once SSD prices come down, they're pretty much already set up to replace HDDs.

schmidtbag:

I agree but drives like this ain't the reason. Unless they do something to improve sustained read/write speeds (even sequential), drives like this are going to be too impractical due to taking too long in the event of restoration from disk failure. Being mechanical (and a Seagate...), the probability of drive failure is high enough that it might make sense to just buy a smaller drive, possibly even an SSD. While there are use-cases for these drives, they're rather limited. Seems to me they're best used for archives. As for NAND, because it's just a chip on a PCB, it can take many form factors. So even if the data density isn't as great, there's already 100TB drives in a 3.5 inch form factor, from ExaDrive. The problem is of course pricing. So - once SSD prices come down, they're pretty much already set up to replace HDDs.

Archives and continuous use of a low-level of constant performance, where additional performance offers no benefit, in something like a small NVR or NAS. But even those are for sure being encroached upon.

schmidtbag:

So - once SSD prices come down, they're pretty much already set up to replace HDDs.

That has been the phrase for 10 years now, and ... it's not really happening. Don't forget, corporations still use tape to archive data, to this day. HDDs will stay as nearline storage for many, many years, until some tech better than SSDs appear (Sorry 3DXpoint, you had your shot), that would replace both and be both cheaper and better in every possible way.

wavetrex:

That has been the phrase for 10 years now, and ... it's not really happening. Don't forget, corporations still use tape to archive data, to this day.

No, the phrase for the past 10 years was until price and capacity match. As of a couple years ago, capacity is now superseded. Now all we care about is price, which has been on a steady decline. It's only a matter of time, though it still will take several more years. I predict before 2030.

schmidtbag:

I predict before 2030.

I really wish your prediction would come true. But I predict they won't, and SSD cost per TB will still be much higher than HDD. Let's meet here again in 7 years and see which prophet was right ;-)

I just upgraded from a 4tb storage drive to a Seagate Iron Wolf Pro 16tb last month. Should last quite a few years. It was $269 after tax but amazon had a deal of 5 monthly interest free payment so I couldn't really pass it up. Sure is nice that they have developed new higher capacity storage drives. I don't see myself filling this one but it nice to know that if I ever do there will be larger drives that I can upgrade to if needed.

Monolyth:

Doubt we'll see more then 500MB/s from HAMR 50TB.

Multiple servos controlling more than 1 head group independently will at least double the DTR, this was recently demonstrated. Current 18TB single servo drives break 270MB/s on the platter outside edge, use of 2 or 3 servos will pass 500MB/s. Faster for higher capacity drives.

Mufflore:

Multiple servos controlling more than 1 head group independently will at least double the DTR, this was recently demonstrated. Current 18TB single servo drives break 270MB/s on the platter outside edge, use of 2 or 3 servos will pass 500MB/s. Faster for higher capacity drives.

I don't think you are talking about dual/multi-actuator which already exist (Seagate MACH.2) but maybe something else? If you are talking about something else, I would appreciate a link if you've got one! 🙂 In terms of dual / multi-actuator there are positives and negatives but IMO the positives (increased speed / multiple addressable LUNs per drive) do not outweigh the risks and TCO. The first issue is that you must have hardware & software support to be able to address the drives properly to realize the performance. That means more complexity (and cost) on your controllers. The second is that while MTBF may still be advertised similarly to existing single-actuator designs, you are still doubling the complexity of the drive. Means more cost and more points of failure. Only time will tell if these gain traction. It's neat tech, don't get me wrong, just not something I'd want to invest heavily into without having some real world metrics on failure rates.

There is a concern regarding HAMR when it comes to data recovery. Sometimes, advanced technology undermines a certain tech service area, rendering tools and software obsolete. Actually , that's the "normal" trend and everything has to adapt and evolve. But HAMR is a different thing all together. Let's hope that a solution will be found to work with it.

I wasn't talking about dual-actuator drives, but simply about the fact that heated platter can be packed a lot denser than non-heated one. The reason is the write and read coils are normally massively different in size, the write one is a lot bigger and stronger to overcome the magnetic resistance of the medium and change the polarity of that particular spot. Being big and bulky, results it cannot very precisely target a certain area, so the bits are much larger than they really need to be. By temporarily heating up the medium at that particular spot, it makes the magnet a non-magnet for a split second, and the write coil can be made much smaller and realign the iron particles with a fraction of the energy. That means, the bits themselves can be a lot smaller, increasing potential density by orders of magnitude. My guess is this tech took so many years to perfect because of the extremely precise positioning and timing of both the laser and the read/write coils, as even the slightest nano-second error would result in heating up and cooling too early or too late, degrading the signal strength and making the platter unreadable. Also smaller bits means even more precision in the mechanics, which is definitely not easy to do. @anticupidon reading will probably be as easy as it ever was. It's the writing that is always tough to do... So not affecting data recover much, their tools just need to get a bit more precise.

I tend to start with one, main drive that holds all of my storage... Then I outgrow it and add another, larger drive.... Then I buy a huge drive and merge it all back into one.... lather, rinse, repeat. I wonder if it's nearing time to merge back to one huge drive again.

Monolyth:

I don't think you are talking about dual/multi-actuator which already exist (Seagate MACH.2) but maybe something else?

This was what I intended but read about it from another perspective, cant remember where. There was suggestion it will work its way into consumer drives but I'm a bit vague on it now.

Demise of the HDD? Even tapes are still out there being used for economic archival. Everything has its use case.