Slashdot Mirror
The Billion-Dollar Bet on the Future of Magnetic Storage (ieee.org)
For several decades, the areal density of hard disks increased by an average of nearly 40 percent each year. But in recent years, that rate has slowed to around 10 percent. Seagate and Western Digital, the leading manufacturers of hard drives, differ with each other on how to get around this. From a report: In back-to-back announcements in October 2017, Western Digital pledged to begin shipping drives based on what is known as microwave-assisted magnetic recording (MAMR) in 2019, and Seagate said it would have drives that incorporate heat-assisted magnetic recording (HAMR) on the market by 2020. If one company's solution proves superior, it will reshape a US $24 billion industry and set the course for a decade of advances in magnetic storage. Companies that wish to store huge amounts of data do have other options, but hard drives are still the go-to choice for enterprise storage needs that fall somewhere between faster, more expensive solid-state drives built on flash memory, and slower, cheaper magnetic tape.
Seagate now aims to debut a 20+ terabyte drive based on HAMR in 2020, and Western Digital promises MAMR drives that will hold roughly 16 TB later this year. Western Digital expects to quickly scale up to MAMR drives with 40 TB of capacity by 2025, while Seagate believes it can achieve similar capacities through HAMR, though it has not publicly stated a target date. Both companies are essentially starting from the same place, with hard drives that share a few key components. The disk, for example, is a thin platter that has been coated with some form of magnetic material made up of countless individual grains, each of which is magnetized in one particular direction. Ten or so grains in a cluster, all with magnetization pointing in the same direction, represent a bit.
Seagate now aims to debut a 20+ terabyte drive based on HAMR in 2020, and Western Digital promises MAMR drives that will hold roughly 16 TB later this year. Western Digital expects to quickly scale up to MAMR drives with 40 TB of capacity by 2025, while Seagate believes it can achieve similar capacities through HAMR, though it has not publicly stated a target date. Both companies are essentially starting from the same place, with hard drives that share a few key components. The disk, for example, is a thin platter that has been coated with some form of magnetic material made up of countless individual grains, each of which is magnetized in one particular direction. Ten or so grains in a cluster, all with magnetization pointing in the same direction, represent a bit.
Once they get down to about $100, I guess a few drives would be enough to backup all the devices in my home, and maybe store some other media.
Until then rsync diffs to a remote NAS will be good enough.
There isn't much on my systems that can't be downloaded again after installing extra apps and my personal data.
Thermal cycling is known to weaken materials (not good for things spinning at a high speed for long periods of time,) while microwave emission is generally hard to control in a cheap and precise way.
Every time I see posts about hard drives getting bigger, I wonder: how long until they're no longer practical due to concerns about data safety? Backing up a large drive is already difficult.
Then again, I would really like to see them make this kind of progress with SSD... A 10TB SSD would be a wonderful thing. :)
-- sigs cause cancer.
So what does /. think about the below?
I am uneasy about 2TB+ drives. The way I see it, that is a lot of data running on a single/small set of failure points. At about 2TB, that's about all the corporate data I generate over 2-3 years.
Each year I trim out a lot of stuff and zip it to about ~50GB of important, must keep, stuff. With 2TB drives, we tend to just keep everything.
And I just feel more is at risk with few protections. One stolen laptop, bad disk jolt, header jitter, etc and so much is gone. I just feel one big drive is so much more risker than multiple drives.
And network backups don't really protect enough because the communications channels are so slow. Most still operate at 50-100mbps. Restoring a lost drive takes forever and that long single write run isn't something these drives are designed for.
And even in servers, isn't 6 raided drives better than 2? Thoughts?
Have you bought an SSD lately? They are mostly air as it is. There is absolutely no reason these couldn't be packed with newer chips to the same degrees as the solid bricks these drives were a few years back to make 200+TB SSDs. There is no particular reason that we need magnetic drives or similar capacity SSDs should cost significantly more. The drive manufacturers just have a common interest in maximizing return on every bit of infrastructure they own and have formed a consensus around it.
This isn't much different than the Telcos continuing to sell "T1" lines with nonsense about SLAs and guaranteed bandwidth. It had very little to do with offering the best product they could and almost everything to do with maximizing margins at consumer expense.
I'm not saying it isn't good for their businesses but when it comes to technology, artificial scarcity like this impacts R&D and availability of products and services for everyone else in a very pronounced negative way.
Tech companies (and others) are forming virtual international monopolies because they actually understand the work of John Nash and that there is more profit to be had by collaborating. This simply does too much damage for the rest of us to allow in certain key areas like Food production, healthcare, energy, defense, communications and especially technology.
Most free market economists do not account for the work of Nash. The entire concept of an unrestricted market which naturally organizes due to competition and market pressures breaks in a very serious way when you can prove that it can be more profitable to cycle non-loyal consumers between a small number of competitors than to actually compete with them in areas in which would reduce margins for all competitors.
It's HAMR. HAMR will beat MAMR.
I don't mean that HAMR will succeed -- it might not come to anything, and/or some new thing might appear that is even more successful -- but between the two, it's HAMR over MAMR.
Because it's not mostly about manufacturing costs or speed or reliability. It's about sales. Guys will buy HAMMER tech and avoid the clearly breast-referencing MAMR, and non-tech folks are NOT going to want obviously cancer-causing microwaves in their laptop.
It's not about logic, it's not about technical merit, it's obvious which one can sell and which one cannot.
They should rename MAMR Wave Assist Recording, because WAR would stand a marketing chance against HAMR.
Is no one working on optical storage anymore? I would think there are more opportunities to increase speed/density there than with magnetic storage.
Support Right To Repair Legislation.
Yep. SSDs are currently about four times the price of HDDs per byte. That's a massive drop over the last few years and the difference is only going to get smaller with time.
These new HHDs will have to be really cheap to keep HDD alive for more than a year or two. If they cost more than current HDDs then there won't be much point to them.
No sig today...
"640K ought to be enough for anybody."
Or in the case of hard disks, a few terabytes.
I'm actually semi-serious: it seems to me that the days of mechanical storage are numbered. With SSDs, and now Intel's XPoint, one can seriously hope that hard disks will be phased out just as floppies were. Fewer mechanical parts ought to mean more reliability, not to mention the obvious speed advantages. Granted, I did buy two hard-disks last year, but only to replace disks in an existing NAS. Those might well be the last ones I ever buy...
Enjoy life! This is not a dress rehearsal.
Don't worry, software will bloat along at the same pace!
Every time I see posts about hard drives getting bigger, I wonder: how long until they're no longer practical due to concerns about data safety? Backing up a large drive is already difficult.
Backing up a large drive has ALWAYS been difficult. The only thing that changes is the size of the number. Some of my early machines have 40MB hard drives and I had no practical means to back up that much data at the time. Now it might be 40TB but the problem is the same and so are many of the solutions. Back then we had tape, second hard drives, removable discs. Today we have... tape, hard drives and removable disks (solid state or optical instead of floppies). The more things change the more they stay the same.
Then again, I would really like to see them make this kind of progress with SSD... A 10TB SSD would be a wonderful thing. :)
How about 30TB?
ceph with more smaller disks can be better then 3-6 super big disks with high rebuild times.
I suppose the you've never needed or used a NAS then. You can use SSDs, it'll just be way more expensive and less reliable.
Well, there's spam egg sausage and spam, that's not got much spam in it.
need more pci-e lanes / bigger pch link to make a few sdd really not get speed capped.
Even a few sata ones can over an pch link and / or an SAS back plane
I would have to say that HDDs are far more reliable than SSDs and probably why they will be in the enterprise a bit longer.
Well, there's spam egg sausage and spam, that's not got much spam in it.
The idea that tape is somehow cheaper than spinning disks is an idea that dates back to the last century. Yes the individual tape is cheaper than a commensurate amount of disk storage but by the time you add in the drive cost, the robot cost, the software cost, the recovery cost (to hard drives) and the electricity then suddenly the tape cost is now greater than the cost of the disk. Even if you take the tapes out of the robot and store them on a shelf the cost today of hard drives is cheaper than tape.
The major selling point of optical was cheap distribution of read-only data like music and video CDs, DVDs, and Blue-Ray.
Streaming makes this much less important. Sure, it would be ncie to have a consumer-priced "super blue ray" reader that could store a full-length 3D 8K movie but when most people would rather stream it, do I really want to spend the money to develop such a device?
Yes, there are still two important reasons for optical media that will keep the market alive for at least another decade or two, but they aren't the "huge" market that drives fast innovation:
* Video collectors, who still want a "factory made" medium like a DVD to put on their bookshelves. For music, Vinyl serves a similar purpose.
* Archivist, who need the long-term storage provided by "1000 year" (note the quotation marks) metallic-dye optical media that will probably last at least 50 years under archival conditions.
Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
If you work in a datacenter I'm guessing this is good news everyone!, but my own storage needs have (somehow, for completely unrelated reasons *cough*) gone down since the advent of streaming services such as Netflix.
Sure, games are getting bigger but I'm not a teenager anymore, so I buy maybe a dozen games per year at the most. Last year I only bought seven and that number is inflated because I bought a cheap bundle of five games on Steam.
#DeleteFacebook
Ok, yes if you save it in the cloud then you data will at some point will be save on some sort of storage medium, most likely a magnetic hard drive(s).
However for normal consumer usage. 1tb is more then enough, and it has been that way for a long time, because most of the data that we consume is on the cloud and in general while it is on the cloud the data is more optimized. For example, if you are to backup all your applications, on the cloud normally there would be one copy for thousands of users, and redundant data would be better optimized, such as with a differential data backup vs what we did in the past is a full backup even if you didn't change anything in years.
This is part of the reason why Cloud services can give you these big storage amounts for much less then you can get your own drive.
The companies putting money in magnetic storage, I hope they are not targeting the consumer market for those. Because I think the heyday of selling big drives to everyone is over, especially when people get a new system with a fast NVME drive. Because having a fast snappy PC is much more important for many people then just having gigs of data available but not touched.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
Just look at Backblaze's regular storage updates. They clearly show that WD and Seagate have much, much lower quality on their drives than Toshiba and Hitachi. As with many other things, the American brands are more popular because of aggressive advertising, but hold much lower quality than, typically, the Japanese brands.
If you still need HDDs, you will do yourself a favor by buying Toshiba and Hitachi (or HGST).
Sure, the performance would hit bottlenecks (does now, it isn't like there aren't SANs populated with SSD now)... but how is that not a better problem to have than having slow and low capacity magnetic storage? Solid state scales to well over 200TB per disk with existing technology if they package it that way and the silicon isn't any more expensive than in the past when the chips were lower capacity, it is just more efficiently utilized.
I'm speaking to a technical crowd, so technically yes, some of the newer fab technologies require longer exposures and the like but nothing that can even begin to explain the small capacities and massive premium they are charging for these drives. They used to largely use the premium prices to pay for the R&D and new equipment and then drop prices. Now they ride the higher prices as long as they can and use articles like this to misdirect people... the next high capacity low cost storage technology already exists, they just don't want to race to the bottom.
This is a great idea. Exactly the same width as the SATA+POWER connector -- 4.5cm. Honestly it doesn't buy you much space, but you can lock out the platter drives. The companies which ONLY make SSDs should definitely be doing this.
Put a couple of notches into them, so they snap in -- no vibration issues demanding screws. You don't need a metal frame around them. Hot-swappable in a sexy way without having to have expensive extra carriers. They should have a standardized hole in the plastic at the front to allow you to snap in a tab/handle for easy pulling.
They'd fit in little carriers (Wee! Something else to sell!) to make them fit 3.5 inch screw-in spaces, and make people think 3.5" is an archaic form, like the 5.25" adapters for 3.5" drives did.
This would establish the new standard without alienating anyone, and basically put the platter companies out of business. No one is going to retool to make even smaller platters.
Oh, wait. Did you mean that the platter companies should start selling 5.25" drives again? Oh, that's is so not gonna work.
Enterprise is ditching HDDs in favour of SSDs as fast as possible for everything except bulk, low speed storage. The performance gap between SSDs and HDDs is vast, and RAM for cache is relatively expensive per GB. For some applications, particularly anything database related (including mail servers, often one of the biggest and most business critical operations) SSDs are impossible to beat.
Reliability isn't a major issue, accounted for with RAID and backups.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
Hammer
Mammer (ies)
Makes me think of Thor's hammer, vs nice tits
"Western Digital promises MAMR drives that will hold roughly 16 TB later this year", Western Digital has been so far behind in delivering things it promises I wouldn't count on this in any way. Seagate has been shipping 8tb desktop drives for a while now and Western Digital still doesn't have a 8TB Blue or Black drive listed on their website.
HAMR, from the name alone scares me a little, knowing nothing of the technology. And, I'm sure I could do some research and will wait for reviews, but initial impressions do take a toll.
My concern would be that cooling, and managing excessive heat inside a small case is already a major problem. So, how they generate the heat? how do they prevent heat leakage or handle cooling? these would be some of my concerns as a hobbyist picking a new drive;
Though, these might not ever be targeted to the home user.
No we aren't. As somebody that works in the enterprise space (SAP), I can tell you we aren't dropping HDDs. We have partnerships with cloud providers Microsoft, Amazon, and Google for our Hana database, and they are not switching to SSDs. Enterprise databases relay more on memory that storage for speed. We spec our servers with as much memory as possible with 6TB as the starting point. HDDs are not going anywhere because SSDs fill a niche role with their limitations.
"And no, it's not the cost of sand, idiot."
You might have got me there if I'd said it was the cost of sand.
"Creating silicon wafers isn't free"
That is certainly true and while the heavy energy input is part of it, most if it just that the equipment is made in small quantities with ridiculously high per unit cost.
"The price on SSDs reflect a global game of chicken, as each vendor is trying to maximize the price while still undercutting their competition."
Yet they publish roadmaps far in advance. You'd never do that unless you and the competition are keeping pace with each other within a reasonable margin rather every release being your best shots. AMD has made Intel shave a year off its planned releases by jumping ahead but the thing is, Intel only had to scramble because it wasn't targeting making its next release the best it could manage, it was targeting making it keep pace. The new CEO broke the unspoken agreement. The same keeping pace gentlemans agreement is in place here. The pricing tiers are more or less fixed within a certain margin and capacity is grown as slowly as they can make plausible. All the manufacturers have this common interest, if capacity grows too much too fast people don't need as many drives, reliability increases (fewer drives in your array means lower probability of failure), margins shrink, etc.
It's no different than cell phones. You ever wonder why all the providers have the same backwards concept of giving steep discounts to new customers but not on upgrades? They actually make more money if people keep switching, dropping grandfathered older plans, and buying new phones. They could build their model around retention but then they'd be in serious competition with the other providers, this way if you are sprint and 80% of new contracts are going to verizon it is no big deal because next week 80% will be coming to you. You just keep cycling around the contracts and the size of the pie is grown so much all the providers win. Creating churn and competing in inches can far more profitable than lapping the competition not mention carries far lower risk of running afoul of anti-trust concerns.
Great news for me....I'm a media whore. My home fileserver has 40tb of space and it's full, spread across 20 or so hard drives of varying size, age and brand.
I do not belong to the church of the lowercase 'i'
I suspect the distinction is simply to allow one company to get around the other company's patents. HDD areal density is currently at about 1 Gb per square inch, or 155 Mb per cm^2. Each bit thus has an area of about 800x800 nm.
Microwaves have a wavelength of roughly 12 cm, which is 150,000 times bigger than 800nm. There's no way you could aim microwaves precisely enough to heat up the surface area that represents a single bit on a disk platter. Both HAMR and MAMR probably just rely on injecting a small amount of heat quickly enough that only the platter surface closest to the heating element experiences substantial temperature increase.
Beta was far superior to VHS. Guess who won that battle. L-1011 vs. DC-10, ugh! Then we have Mac vs. PC, uh oh!
“He’s not deformed, he’s just drunk!”
My hard drive is already large enough, I just want it to last longer.
Semiconductor manufacturing is different to cell phone manufacturing. The cost of entry is orders of magnitude more expensive and each new generation has an investment cost that makes nuclear power look easy. You can count the number of contenders in the market now on one hand because of this. Intel having trouble moving on to another generation shows how difficult it is.
Facts are history now plebs have politics for religion on social media.
In the paragraph "It's no different than cell phones" I was referring to the collaborative/competitive dynamic between the existing players and not the engineering. By syncing up for the most part and competing on marginal adjustments they make more profit than by releasing a market upset increment to try to steal the market outright.
Although, for that side of it I could have said "it is little different than the chips in cell phones."
I absolutely do not dispute the need for very deep pockets to enter this field. But there are people who have them and also have primary businesses that could make enough use of the output to justify that outlay, even with a break-even or loss leader target for that particular segment. There are already some tech companies inching in this direction for certain needs already.
A new player would have some big advantages. They can utilize technology that lends to mass production but isn't necessary consistent with current process lines and equipment. They build a structure from the ground up to target commodity supply and never become dependent on revenue streams which depend on overcharging at the high end or classifying flexible features as "enterprise."
There are mega wealthy billionaires who are blowing obscene amounts of money just to play with rockets and there are no shortage of people in that category with tech dependent business. Is it really so far fetched that one might use that economic power to liberate chip fabrication?
Consumer NAND is only 4x the price. Professional-grade/enterprise SSD are still up there at 10-20x the cost of spinning rust. Even at 4x the price, you're still looking at an average investment of $1M vs $250k.
Custom electronics and digital signage for your business: www.evcircuits.com
"Well, actually."
The problem is that where SSDs are used they have significantly less lifespan because they are read/written so often. For example they are not used in long term storage because the cost per GB is so much higher.
Well, there's spam egg sausage and spam, that's not got much spam in it.
I don't know about "ditching". There are scenarios where Enterprises may use SSDs but there are still many use cases where there is little advantage. For example in the consumer space the main advantage is much faster read/write speeds for applications as consumers open and close applications all the time. In the server space, many applications are loaded on startup and occasionally may be rebooted. However rebooting an application is generally the exception. Putting in a SSD will only marginally improve performance especially for the cost difference.
Well, there's spam egg sausage and spam, that's not got much spam in it.