HDD to stay relevant for the next 15-20 years, SEAGATE CFO said

[zMeul]

source: http://www.anandtech.com/show/9858/seagate-hard-disk-drives-set-to-stay-relevant-for-20-years

at the Nasdaq 33rd Investor Program Conference, David Morton (chief financial officer of Seagate) said:

I believe HDDs will be along around for at least 15 years to 20 years

The very first hard disk drives (HDDs) were demonstrated by IBM back in 1956 and by the early 1980s they became the dominant storage technology for all types of computers. Some say, hard drives are no longer relevant as solid-state drives offer higher performance.

yes, they do but ...

SSDs' price per GB is still a lot higher than compared to a HDD - yes, prices for SSDs will go down, HDD prices will do the same

CDs appeared in 1982 and became obsolete when DVD format was commercially available circa 1996

look at floppy disks - were commercially available in 1971

HDDs are here to stay - they already have 35y under the belt and will stay with us for at least one decade more

sales of HDDs is decreasing, but technology keeps evolving

[Xenift]

In a way, HDD is CD.

We are still using technology from 40 years ago so heavily, not like we have flown to another planet to live just yet.

[TheRandomness]

HDDs will still be relevant because have you seen the price of 8TB SSDs?

[patrickjp93]

He's right, and frankly HDDs can catch up to SATA SSD performance by making the arms asynchronous, vastly improving random read and write. Heck they can even vastly improve sequential that way too by striping the data across platters. Seek time will still be an issue, but you can't win at everything.

[ThomasD]

... by making the arms asynchronous...

Much easier said than done.  Also such a change would make those cheap spinning platters much more expensive to manufacture, and therefore to purchase.  

 

Which, as SSD prices continue to drop, would only hasten their demise.

 

No, the only hope HDD manufacturers have is to increase capacity and lower prices, but even that is only delaying the inevitable.

[Stefan1024]

Just to remind you, tape based storage is still used i data centers for "cold storage". So HDDs will still be used for decades.
Only if SSDs are better in (all of them)
- cost / Gb
- volume / Gb
- energy / Gb
- livetime
they may kill the HDDs eventually.

That said in the consumer market this may happen earlier.

[MrDynamicMan]

I see SSDs being dominant as operational drives in the next 4-7 years. Hdds will become mainly archeivel and tape will remain as cold storage.

[AresKrieger]

Depends on what they manage to do in terms of improvement to the current tech, if they actually achieve their crazy goals, then I'd agree with him, but if not I'd wager 10 years max in the consumer market.

[imthewalrus]

If Intel's 3D XPoint claims are true, then that is going to be a lot sooner.

[That Norwegian Guy]

HDDs will still be relevant because have you seen the price of 8TB SSDs?

 

They are cold storage HDDs, that's why they're cheap. If you buy them to be your active media drive, it'll croak in 9 months and torture you with noise for the entirety of that time.

Edit: I thought you were talking about HDDs there, nvm.

 

But still, HDD will remain relevant, but I don't know about 15-20 years. My personal guess is 10-12 years.

[qwertywarrior]

Tape will stay for a long time too

but eventually SSDs will over take HDD in every way

[patrickjp93]

Much easier said than done.  Also such a change would make those cheap spinning platters much more expensive to manufacture, and therefore to purchase.  

 

Which, as SSD prices continue to drop, would only hasten their demise.

 

No, the only hope HDD manufacturers have is to increase capacity and lower prices, but even that is only delaying the inevitable.

 

The tech was demonstrated a long time ago. It's not as hard as people make it out to be. And the added expense would be minimal. Sure a new controller would have to be devised, but SSD makers did the same thing with the change to SATA Express and NVMe.

 

SSD prices won't continue dropping. As lithography tightens, endurance takes a big hit. It will require new materials to keep scaling SSDs as we have. As much as Seagate stuck its foot in its mouth about SSDs remaining niche about a year ago, the CFO isn't wrong about this. There's plenty of room on HDDs to innovate both in density and in performance that doesn't rely on space-age materials.

 

Increasing density is of no issue, especially once they start using both sides of their platters.

[patrickjp93]

A conservative statement.

 

But as far as consumer storage (where most of the money is at) is concerned, HDDs have no place to go in the long run.

 

One day miniaturization will give us handheld devices with the power of modern desktops. The storage solution that people need will have to be small, use negligible power (not enough to spin platters), and shock proof. How do today's smartphones compare to the the best consumer desktops 15 years ago?

 

Yes they can both drive down the cost/GB, but how much do you think the raw material cost compares between a tiny USB drive and a HDD?

Price of silicon vs. magnetized iron oxide? The HDD wins in being cheap in BOM.

[vanished]

Of course seagate would put this out.  I bet WD's happier and happier about their SanDisk deal every day

 

That said, I agree with the article - they probably will be around for another 20 years.  Just not in my PC  I think consumers will be off them entirely within 10 years, even if SSDs do still cost more; it will be worth all the other advantages, even if HDDs somehow catch up speed-wise*.

 

* Which I think is questionable; they could probably improve to match SATA SSDs now, but what will the common SSD be like by then?

[Gibs960]

IF SSDs were even slightly more expensive for the same storage, I think HDDs will die pretty quick. But I reckon we're a good few years away from that. 

[AlwaysFSX]

He's right, and frankly HDDs can catch up to SATA SSD performance by making the arms asynchronous, vastly improving random read and write. Heck they can even vastly improve sequential that way too by striping the data across platters. Seek time will still be an issue, but you can't win at everything.

If they can get the speeds even half way to NVMe across the performance board it's already a lot better than what everyone uses right now which is a significant improvement.

[Guest Kloaked]

Surprised that TerryTek hasn't posted some dank meme to show his love for HDDs.

[patrickjp93]

I'm talking all material required to make a complete product, not just the storage medium.

There's a lot more than just iron oxide in a HDD.

 

That's the bulk of it. The arms are a pair of Titanium rods with an aluminum wing and a couple copper wires each. The servos for the arms have gotten incredibly cheap too. The motors are nothing special either.

 

If they can get the speeds even half way to NVMe across the performance board it's already a lot better than what everyone uses right now which is a significant improvement.

Currently HDDs do not strip data across platters and all arms on a single column move together. Top R/W speeds right now are what, 210 MB/s for 5-platter drives ala Western Digital Black? http://www.storagereview.com/wd_black_6tb_hdd_review

 

Make the single column asynchronous and stripe data and guess what? You can improve that by a factor of 5 for 1TB/s R/W, and that's not purely theoretical either.

 

Latency will always be an issue, but bandwidth on HDDs is easily capable of exceeding the best of current NVMe SSDs. Most modern drives even come with 2 or more arm columns, so this really should come as no surprise. You'd need a controller to make sure the data was correctly ordered, but that's fairly simple since CPUs have had Out of Order Processing engines forever. The tech for this case is much easier to build.

 

HDDs have plenty of room to make a big leap in performance. Seagate and WD will both remain very relevant for another 10+ years easy, especially since SSDs are going to level off in another 3 years for price/GB once lithographies have to drop to 20nm again to compensate for price scaling, which will bring back endurance problems.

 

NAND flash will die long before spinning rust does. We'll have to move to 3DXPoint, Memristors, or other PCM-based memories to keep "solid state" moving toward being competitive with HDDs.

 

And as per consumers not really needing the capacities given by modern high-end HDDs, as big as drivers alone are getting, and as big as games currently are, I'm afraid that need will soon come back to haunt SSD producers.

[ThomasD]

I'm talking all material required to make a complete product, not just the storage medium.

There's a lot more than just iron oxide in a HDD.

Long term the biggest limiting factor for HDD technology is moving parts.

 

1.  They break

2.  They consume (relatively) large amounts of energy 

[patrickjp93]

Long term the biggest limiting factor for HDD technology is moving parts.

 

1.  They break

2.  They consume (relatively) large amounts of energy 

 

With the advent of NVMe, high-performance SSDs consumer just as much power as a WD Black. Those controllers get quite hot, which is why the 950 Pro from Samsung thermal throttles in a constrained environment.

[lx_srs]

They'll stay relevant for mass storage for a long time, but if you need anything for speed you can forget about getting an HDD.  Also you can bet corporations will keep the IT budget low, forcing the use of HDDs until SSD's are cheaper.

[ivan134]

I feel like this is the 3rd time he has said the same thing this year alone.

[WereCat]

I think that HDDs will still be with us for a long time but only for archive purposes. SSDs will dominate consumer market and most people will have HDD just for backups.. .(it is already happening).

[patrickjp93]

They'll stay relevant for mass storage for a long time, but if you need anything for speed you can forget about getting an HDD.  Also you can bet corporations will keep the IT budget low, forcing the use of HDDs until SSD's are cheaper.

 

I wouldn't say speed = forget HDDs. It means forget current HDDs. There's a fairly simple innovation that will more than saturate the SATA 6Gbps bus and will even pull into the PCIe performance range. It's not yet needed since SSD storage density is nowhere close to matching HDDs. Deploying asynchronous arms will rebalance the scales very much in favor of HDDs' longevity until much newer solid state techniques than NAND flash come down in price, such as 3DXPoint, which is starting at a price medium between DRAM and NAND Flash.

 

Not to mention HDDs have plenty of room for density improvements that don't require the use of Helium to increase platter count. And we haven't even started doing 2-sided platters yet. Between that and HAMR, 40TB HDDs will be available before the end of the 2020s at very comparable prices to current 6TB HDDs, but with far better performance. NAND Flash will never keep up with that.

[jagdtigger]

And we haven't even started doing 2-sided platters yet.

 

As far as I know we use both side of the platters since "ancient times"... :huh:

Edited December 18, 2015 by jagdtigger