Slashdot Mirror
How Intel and Micron May Finally Kill the Hard Disk Drive
itwbennett writes: For too long, it looked like SSD capacity would always lag well behind hard disk drives, which were pushing into the 6TB and 8TB territory while SSDs were primarily 256GB to 512GB. That seems to be ending. In September, Samsung announced a 3.2TB SSD drive. And during an investor webcast last week, Intel announced it will begin offering 3D NAND drives in the second half of next year as part of its joint flash venture with Micron. Meanwhile, hard drive technology has hit the wall in many ways. They can't really spin the drives faster than 7,200 RPM without increasing heat and the rate of failure. All hard drives have now is the capacity argument; speed is all gone. Oh, and price. We'll have to wait and see on that.
Well, the Samsung 3.2 TB drive claims that you can read/write the entire drive every day for five years before failure. It's my understanding that at one point, SSDs were notorious for gradually declining over time, but that today's generation of SSDs basically has reliability out the wazoo. I can't quote you stats on it, but anecdotally, I've had a couple of SSDs in my computer for several years now, I leave it on 24x7, and I've never had a problem.
...Yet. YMMV.
Yes, but they employ alot of techniques to mitigate this. The endurance is so high that unless you are recording audio/video almost constantly, it will usually not become an issue. There's plenty of literature on it so not going to be redundant.
If the price for GB is below $0.50 then they got a winner.
I don't know why Intel and Micron get any special consideration given that right in the summary the fact that Samsung has already announced the same move.
Also incorrect assertion that drives don't go faster than 7200 (there are 15k drives, just they are pointless for most with SSD caching strategies available).
XML is like violence. If it doesn't solve the problem, use more.
It's only wacky if you bet on it happening next year. But, soon enough.
You mean the click of death when the hard drive fails? That I don't miss.
Actually, one of the nice things about SSDs is that as capacity increases, reliability increases too. More cells means more options for wear levelling, means more life span.
That's the thing. Most people who need 3.2 TB of space will only write to each location a few times, and data won't change very often. Sure, some writes will be happening, but not even close to the magnitude that you'd need to wear out one of these drives. There might be some cases in commercial applications where you'd need to write that every day, but the typical desktop or laptop is never going to see that kind of usage.
Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
According to Techreport, Intel's three-dimensional NAND. will enable 10TB flash drives in servers in 2 years
And 15 years ago I bought a massive 2GB drive for $350.
Computer stuff gets cheaper over time. There's no reason the same won't be true for SSDs. At some point SSDs will be cheap enough that even if HDD are still 1/100th of the price, SSDs will still win because of all their other advantages.
> There's no way that SSD is gonna come even remotely close in price.
In which timeline ? I've seen USB 2.0 128GB flash drive offered for 10$ for black friday. That`s 3.8TB of flash memory and tons of redundant parts for 300$. I gather it will not be `too long` before it's not worth it anymore to buy spinning drives.
6TB drives can be had for $250-300
That's really nice I agree, I have a few 4TB drives that I use for photography... but I would without hesitation pay 4x the price for the speed of an SSD (especially one not bound by SATA speeds, like the Samsung PCiE SSD...)
"There is more worth loving than we have strength to love." - Brian Jay Stanley
In theory yes, in practice it's unlikely to ever come up. Wear leveling does wonders, over provisioning does more on top.
If SSDs had come first you'd be saying the same thing about HDDs: Don't HDDs have fragile mechanical parts that fail randomly?
...as long as high capacity SSDs keep costing as much as an entire computer.
BeauHD. Worst editor since kdawson.
And 15 years ago I bought a massive 2GB drive for $350.
Then you got ripped off. Seagate sold 28 GB HDDs for $350 in 2000.
And 15 years ago I bought a massive 2GB drive for $350.
4 years ago I bought a few 2TB drives for $69 each.. Then the prices sky rocketed and still haven't come back down to that price.
I believe what they do is spread the data all over the memory, to mitigate the issue of a small part of the memory being heavily bombarded w/ writes while 90% of it never gets touched. I'd imagine that Copy-on-Write filesystems, such as ZFS, would enable one to do it more effectively, since no actual data ever gets deleted, and only the metadata info is changed, and the changed data is written to another portion of disk. If this is done effectively, then the disk utilization is increased, and endurance issues don't come into play at all.
Otherwise, you are right - cell design does seem to be hitting a wall, and I don't see silicon getting much smaller. Certainly not for price decreases. Also, multiple bits per cell don't lend themselves to too many write cycles, being as unstable as they are.
So I'm dreaming about the racks of 15k drives I have?
Watch this Heartland Institute video
SSDs will likely get there in 3-5 years by Moore's law. The question is where hard drives will be by then.
Effective SSD lifespan goes up with size due to wear leveling techniques. If you manage to write 100 GiB per day every day to a 1 TB Samsung 840 EVO (which uses low endurance memory to get better speeds at its target price point), it would take you over 30 years to physically wear out the NAND cells. If you somehow manage to overwrite the entire drive every day it'll take 3 years on average to break it, the Pro models have tougher memory and would last upwards of 15 years constantly overwriting the entire thing.
To add to my previous post, going even farther back Seagate in 1998 sold a 6.4 GB HDD for $350. If you paid $350 for only 2 GB 15 years ago, you got royally boned.
Okay, 3 years ago, a 256GB SSD cost $900.
Today, you can get them for $100-200.
Chas - The one, the only.
THANK GOD!!!
It really depends what you are going to use it for. If it's your desktop PC, consumer grade drives are fine. If you are going to use the SSDs for scratch storage on a supercomputer or the journal devices for Ceph, you probably are going to want high write endurance drives.
Newegg has 2TB HDDs refurbished for $69. New ones at $79 or higher. Even the 3TB for $105 they sell is actually about the same $/GB as those $69 2TB HDDs.
And 15 years ago I bought a massive 2GB drive for $350.
Computer stuff gets cheaper over time. There's no reason the same won't be true for SSDs. At some point SSDs will be cheap enough that even if HDD are still 1/100th of the price, SSDs will still win because of all their other advantages.
I agree, eventually SSDs will become cheap enough that it won't be worth it to manufacture spinning hard-drives anymore. It's kinda like Plasma TVs today. They are being dropped by TV manufacturers because it's cheaper to scale up LED TVs.
That being said, it's not going to happen overnight. The drive manufacturers need to make their R&D money back, at the very least....
Typically, the endurance of any non-volatile memory (read flash/hard drives) is measured per sector/block, where the latter is the smallest number of erasable bytes/words/quad-words that an erase operation can erase. Typically, for flash, that number is 1-10 thousand cycles. That number is eroded as one increases the number of bits per cell.
Like I mention below in a response to the GP, if you have it so that every byte is written only once and any overwrites happen to other bytes/sectors, you can avoid multiple erase cycles and thereby maximize the life of such an SSD.
And they had 5 years warranty. Today it's 1-2.
Comparing a $10 USB stick with an SSD is like comparing turtles to cheetahs. Those USB sticks might write at 2-5 megs/sec. Maybe. 1/100 the speed of a good SSD. It's not a cromulent comparison.
RAID doesn't protect against loss of data, that's what backup is for. RAID protects against loss of uptime.
Sandforce controllers also do encryption, and certain controllers with certain operating systems can leverage this to integrate the controller-level encryption with the OS-level encryption, at which point the drive compression is done on the raw data before encryption happens.
But make sure to put the SSDs into the RAID a few months apart -- it would be embarrassing to have them all fail at the same moment.
Watch this Heartland Institute video
Companies still make tape drives. There is no reason to beleive HDDs are going anywhere anytime soon. People have been wrongfully proclaiming the death of the HDDs for most of this decade.
It's not disingenuous at all. It merely demonstrates the primary problem here, namely the price gap. Larger SSD drives are low capacity and expensive. They are priced outside the range of most consumers while also being inferior in terms of bulk storage. A larger SSD is less able to justify it's price premium than a larger HDD.
Even if SSD prices get less ridiculous, chances are that HDD prices/capacity will keep pace and continue to keep HDDs relevant.
A Pirate and a Puritan look the same on a balance sheet.
Seems to me that I bought my first 85 MB HDD for about that much 30 or so years ago....
"I do not agree with what you say, but I will defend to the death your right to say it"
Also, didn't Intel exit the flash market a while back, spinning off its flash division along with ST Micro to Numonyx, which later got acquired by Micron? I thought that the whole idea then was that memory was so unprofitable that it wasn't worth keeping it as an albatross on corporate margins.
Also, memory fabs are different from the ones used for making processors/controllers - it's not like fabs that don't make more Atoms or Celerons will be repurposed for SSDs. So how does it make sense for Intel to get into this? Micron I can understand, since memory is their prime business. But Intel? It makes as much sense for them to be making this as to be in the DRAM market
We'll have to wait and see on that.
What's wrong with you people. We are waiting already for 5+ MORE THAN FIVE fucking years. Still hasn't happened.
1TB HDD - 60-80€, 1TB SSD - >350€.
The problem is that once PC is turned on, there is not much use for the SSD speed. It's not like I'm moving terabytes of data around everyday. And even if I have to, I do not have to wait for it: I simply leave it overnight.
Another problem is that (some) SSD have the nasty habit, once failed, to deny you access to the data at all. I hoped that at least those jackasses would straighten out the SMART support and finally standardize the monitoring parameters. But few moronic manufacturers even proclaimed that their drives are so good that they don't need no stinking SMART support...
All in all, SSDs are developing too fast. And have pretty bad history of firmware bugs. And literally all manufacturers, instead of strengthening their stance of data safety, all like one doubled down on the "oh but look how fast it is!"
P.S. And TRIM support is still in shambles. After all the years, some drives still require a proprietary application/driver installed.
All hope abandon ye who enter here.
Usually USB flash drives have really poor write endurance and reliability in general compared to proper SSDs.
All RAID levels protect against loss of data due to failure of individual drive(s), port(s), or data cable(s).
RAID 0 is not RAID.
RAID is not backup.
Wear leveling does wonders, over provisioning does more on top.
Add compression on top of that. If your data isn't all ZIP, PNG, JPG, MPG, or some other compressed format, the controller turns repetition into even more over-provisioning.
The writer is simply ignoring cost as an inconvenient fact of SSD adoption rates.
I've been using Seagate's hybrids for a couple years and the combination of performance, simplicity, and economy hit the spot. I have 750 gig and 1tb drives in my laptops and a 2tb in my gaming rig. The hybrid drives were a small price bump for a big performance bump. Sure, gigantic SSDs would give me a slight performance boost but it's a big jump in price for a small jump in performance over hybrid.
It doesn't matter what RAID level you use, rm -rf / will still dutifully delete all of your data for you.
Repeat after me, RAID is not a backup.
Game! - Where the stick is mightier than the sword!
Most current MLC (multi-level cell) implementations can sustain anywhere between 1000 and 3000 write/erase cycles per cell. This is better than TLC (triple-level cell, max 1000), but far worse than SLC (single-level cell, max 100k up to a million, depending on the technology).
The problem is the way how flash itself works, and how smart your controller is. Unlike a disk, flash must be erased before writing. And here is where the problem comes: flash data is stored in a page of cells, with typically 8 pages of data per "block". Erasing happens on the block level. So in order to erase a single page of data, you need to erase all 8 pages in a block. Since you need to keep the data of the other 7, you first need to copy that data into another block, erase the original one, write all data back and erase your "tmp" block. The churn on blocks happens a lot faster than what you'd think.
Having that said, for consumer products, MLC or TLC is perfectly fine. For enterprise, not so much.
You'll see that in the price, obviously. TLC is the cheapest, followed by MLC, and the most expensive technology is SLC.
I'm not a complete idiot... Some parts are missing.
Comment removed based on user account deletion
Well, the Samsung 3.2 TB drive claims that you can read/write the entire drive every day for five years before failure.
Such statistics are meaningless in my book. Light bulb manufacturers claim their bulbs will last five years or seven years but when you look at the fine print they say that's given under the idea you're turning the light on, leaving it running for 3 hours, and turning it off once per day -- nobody uses light bulbs like that.
Not if you fill that space up. The excess capacity used for wear leveling only works if there is non reported space used for such or you don't fill the drive all the way up. The minute you fill the drive you are using the flash to it's extent. And putting things like page files on them will increase the wear rate significantly. A smart installation puts the page file/swap space on a magnetic disk and uses the SSD for everything else that isn't doing caching where there are heavy writes.
Until they can solve the wearing problem that gets worse with each die shrink of NAND they will never reach the wear capacity or costs of spinning rust.
Comment removed based on user account deletion
Comparing a $10 USB stick with an SSD is like comparing turtles to cheetahs. Those USB sticks might write at 2-5 megs/sec. Maybe. 1/100 the speed of a good SSD. It's not a cromulent comparison.
That may well be, but have you ever looked at a turtle's drag coefficient?
When our name is on the back of your car, we're behind you all the way!
You said: "The article writer mist be smoking some amazing shit to come to such a wacky claim."
Are you referring to the article summary, or one of the specificly linked articles? Because summary says: "Oh, and price. We'll have to wait and see on that."
So they are not making any claims about price. It seems maybe you are the one smoking too much?
Anyhow, there are only a few niche roles where a desktop needs that much space. Give me a 240 GB SSD with 10 times faster IOPs, 10th of the heat and power consumption, zero noise, and no moving parts. That's plenty.
HDD's still have there place for certain use cases, but SSDs beat them by an order of magnitude on just about every factor except price per gigabyte. $/gb is not as relevant when you realize $150 will get you enough of space on an SSD for most desktop roles, and way more than you need on an HDD.
It won't kill the HDD anytime soon.
SSDs just cost too much compared to same-sized HDD.
a 8TB SSD cost $7k, that's just ridiculous when you considering it cost less than $500 for that much HDD storage.
The facts is, data density on solid state has already surpassed data density on magnetic/spinning media. We already have 4TB SSDs at SanDisk, and 8TB and 16TB drives are on the horizon. I highly doubt you'll see HDD do 16TB in a 2.5" package anytime soon. Also, when you factor in power requirements, the economics start to look very favorable to SSD, not 10K or 15K RPM drives which are used in latency-sensitive applications. And, in another unexpected and non-intuitive twist, SSDs have a much higher MTBF then HDDs, around 2.5M hours vs. 1.1M hours. Wear-out has become a non-issue, even in the harshest environments. There are SSDs that are rated for anything from less than 1 Drive Write Per Day (DWPD) on up to 45 DWPD during the length of warranty (usually between 3 and 5 years). Bottom line, SSDs are a far better economical choice in many cases. Of course, this only matters where economics count. Enthusiasts are already seeing the benefit in lower prices and higher densities, but the OP is not about them. Economics is not about "can the consumer afford it" as much as it is about "does this enable me to lower my TCO in the data center". We are there already.
My USB 2 sticks do something in the range 12-15MB/s
30 * 15MB/s would be 450MB, which is close to an SSD speed. And we don`t know if the bottleneck is the memory, the controller or the bus/drivers. I was mostly showing that it will be possible to reach that price and that the current value of the memory itself is already close while the parent was saying that SSD will never reach price parity with HDD.
I could also have pointed out that in the last 3 years HDD prices have mostly stagnated while SSD prices went down very fast for even better performances. I guess we will see!
that NAND flash prices per megabyte have plummeted while EEPROM prices per kilobyte have remained high and then wondered how that could be rational?
NAND flash was a "gimmick" in which a trade-off was made between density and cost versus quality and lifespan. NAND flash ships all start off with failed bits and all the good bits are on the path to failure from day one. With each write cycle, you risk killing an NAND flash bit, and given that all NAND flash devices are being treated as disk drives, if you kill more bits in a sector than can be corrected by the error correction bits then the entire sector must be marked bad and not re-used.
You can write an app that will write new data over old data on a spot of a hard drive and let it run for YEARS without degrading the drive, but if you run that exact same app on a solid state drive you will, depending on the specific NANDs used and the speed of your code, soon begin to lose bits. As enough of those bits fail, you lose those sectors (though the drive and your OS will shield you from seeing this as the drive slowly just appears to have a lower capacity) and eventually the drive will probably fail as the code in the drive runs out of spare sectors to re-map
If your primary use for a drive is to write huge files infrequently and read data from the drive very frequently, then a NAND-based solid state drive will be your best choice assuming the price per GB is reasonable (this is even more so if you are in an environment where jostling might crash a spinning drive). If, however, you are running software that creates and writes lots of data to the drive (like log files, live data being captured, and so-on) with lots of re-writing activity then motors and a magnetic platter are your friend. Ask yourself how many million write cycles your NAND flash cells can take, and then consider how fast your computer can exectute that many write cycles.... remembering that sometimes people unleash code that has bugs and gets stuck in loops while running unattended....
As with many things in life, selecting the proper tool for the job is important.
Backups aren't the discussion here, data loss due to drive failure is.
Context means everything, and you can't use boring thoughtless mantras to answer every question.
There is storage, and then there is storage. 6TB drives are nice and all, and good for things like Pictures and Movies, where you don't need access speed. The moment you have databases, which need things like IOPS then size doesn't matter nearly as much. When you see the difference in IOPS between Spindle Drives and SSDs, you'll start to realize there is more to storage than size.
Your high end 15K spindle drive can do about 150 IOPS The only way to increase this is to go to RAID. A good High End Spindle RAID can do about 10,000 IOPS, or perhaps a little more.
On the other hand, the Samsung 3.2 TB Drive claims to do 750,000 IOPS, or 75 times more. So, while you can get all sorts of storage using RAID and Spindle drives, the money spent for IOPS is better spent getting SDD and RAM (yes) drives. Hard Drives are the new "Tape Backup System".
Agent K: A *person* is smart. People are dumb, stupid, panicky animals, and you know it.
Two out of three ain't bad.
When all you have is a hammer, every problem starts to look like a thumb.
HDD:
Trashing for few minutes once a day = updatedb
Continuous clicks/ no sound / screeek = failed
Random activity = normal
SSD:
Nothing, can't even get neodymium magnets or nice machined parts if it fails.
That's Adorable.
Agent K: A *person* is smart. People are dumb, stupid, panicky animals, and you know it.
To be fair, SSDs seem to be progressing much faster than HDDs. If HDDs continue their sluggish capacity growth, SSDs will pass them in cost per terabyte in a few years.
That's actually what they do.
1) Select an empty block.
2) Copy the data into ram on the device
3) Write the new physical block
4) Update the virtual/physical block map
5) Mark the old block as empty
If that's the case, then why are they not copying the data to ram contained on the drive itself? Seems like an awful waste of cycles with a relatively simple fix. Is it just a cost issue?
Cost and reliability/latency. If you copy it to RAM and get a power outage, data is gone. So that will ruin your reliability. Which means that you have no choice but to ack the write after it's written to the actual block itself. Which in turn increases the latency between receiving the data and ack'ing it.
I'm not a complete idiot... Some parts are missing.
It is more than just Price per TB. It is speed (IOPS). What good is all the storage space in the world, if it is slow? If it was just Price per TB, Tapes are even higher density/price, however the IOPS are so slow.
If you read the article on the Samsung SSD, you'll realize why they put it on a bus that wasn't SATA III (Not fast enough)
Size, Speed, Price, pick any two.
Agent K: A *person* is smart. People are dumb, stupid, panicky animals, and you know it.
I agree, eventually SSDs will become cheap enough that it won't be worth it to manufacture spinning hard-drives anymore.
Capacity per dollar. Home use, 2TB is fine. But in business, arrays of 50TB are common, and size will only grow. Eventually spinning rust drives will become the near-line storage we used to have when tape and laser disks actually had large capacities.
First HDD that I had in my first PC was 40MB, it was $600 so yeah prices sure have gotten better.
Om, nomnomnom...
Exactly. RAID prevents a drive failure from being an immediate data loss.
Plus, RAID allows me to keep all of my bulk storage online while I am replacing a drive. That portion of my data hoard is not completely unavailable to me while I am copying data to the replacement drive.
Of course you want at least 2 copies of your data.
This is where the relative cheapness of HDD wins the day. You don't just need 1x of what you think you need but at least 2x.
Yes. Take that large number associated with SSD tech and DOUBLE it.
A Pirate and a Puritan look the same on a balance sheet.
Or, if your house burns down
Under common usage patterns, you can expect up to 25 years life for first-gen SLC, 15 years for first-gen MLC, 10 years for current-gen MLC, 5 years for current-gen TLC. MTBF is lower due to defective cells and such, but that's also true for HDDs.
Correct me if I'm wrong, but don't SSD's have a point where they put on too many write's per bit?
Tech Reportchecked a bunch of SSDs for write durability and virtually all of them made it to 600 terabytes of data writes or better.
For an ordinary desktop user, write durability is not a problem. Now what about storage durability? With 3 bits per cell, how long before the data fades?
You're both right. RAID can decrease the chances of data loss due to some kinds of problems, but ultimately it shouldn't be considered a reliable protection against data loss. A RAID can be lost or corrupted, or someone can overwrite or delete a file. If you want to assess the risk to your data and talk about the set of data that is protected against loss, you should only consider your backed up data to be "protected". The protection that RAID offers is too weak to be considered to be significant protection.
Therefore, the fundamental purpose of a RAID is to prevent the downtime due to failure of an individual hard drive. If you did not have RAID, then your data volume would stop running, and you'd have to be offline while you repair the device and restore from backups, so that's what you're successfully preventing. All the data that has been backed up (assuming your backup is good) should be safe, and any data that has not backed up is not safe, regardless of whether you have a RAID.
RAID is redundancy, not backup.
you first need to copy that data into another block, erase the original one, write all data back and erase your "tmp" block. The churn on blocks happens a lot faster than what you'd think.
If that's the case, then why are they not copying the data to ram contained on the drive itself? Seems like an awful waste of cycles with a relatively simple fix. Is it just a cost issue?
Any wear levelling worth its salt will not do what the grandparent wrote. You simply do not change one page in a block. If you write a single page, that is handled by mapping that page to another (free) block and maintaining a mapping table for which LBAs are currently stored in what blocks. However, if you are doing single-sector writes, or in turn repeated I/O flushes of the same sector, you still see a lot of write amplification. To keep data integrity, the mapping tables also need to be kept updated in a correct way (or at least uniquely recoverable by scanning through all blocks after a hard power off).
In 1987, I bought an 80 MEGAbyte drive for $775 (around $1600 today), thinking how amazing it was that disk drives had broken the $10/MB barrier. When the first 1GB drives came out a few years later, I remember thinking, "Who would trust that much data to a single device? What an amazing single point of failure!" Now there are 128GB MicroSD cards for under $1/GB. Even understanding the technology, the mind boggles.
The 'wear out too fast' concept is wildly overblown. You can listen to old rumors, or read actual test data.
600TB total writes - http://techreport.com/review/2...
800TB total writes, and some of these consumer grade drives start to fail - http://techreport.com/review/2...
"By far the most telling takeaway thus far is the fact that all the drives have endured 600TB of writes without dying. That's an awful lot of data—well over 300GB per day for five years—and far more than typical PC users are ever likely to write to their drives. Even the most demanding power users would have a hard time pushing the endurance limits of these SSDs."
By contrast, my main home machine (120GB Kingston SSD) has ~7GB total, in over 2 years of 24/7 use. I'll leave you to do the math on lifespan for that.
Happening a little sooner than I thought, but the trend has clearly been going in this direction for a long time now. Just one year ago I stopped buying 3.5" HDDs a year ago in favor of a combination of (short stroked) 2.5" drives and SSDs. I already use only SSDs in all the workstations and laptops, the HDDs are only used by the servers now.
Now it is looking like I will probably not buy any more HDDs at all, ever again, even for the servers. That is going to do wonders for hardware life and maintenance costs.
It's a bit strange having a pile of brand-new perfectly working 1TB and 2TB 3.5" HDDs still in their static bags, unopened, in my spare drawer that I will likely never use again.
I wonder how long it will take case makers to start giving us 2.5"-only hot swap options without all the 3.5" crap taking up room. Of course, there are some already... I mean for it to become the predominant case style.
-Matt
Most people don't need their drive to be that fast. We'd all like faster storage, but HDDs are so slow that a fairly slow SSD is still pretty speedy.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
The actual wear leveling algorithms are proprietary, but rest assured that they do not use flash as temporary memory, and neither do they read an erase block, change one sector and write the erase block back. One thing flash controllers do is maintain a list of unused sectors. So, if you write to one sector, the data goes into an empty sector of a different erase block and the controller remembers that the sector's old location is now unused (and where the sector is now). That's where the TRIM command helps: It marks sectors as unused without using up a different sector somewhere else. When the drive needs more free erase blocks, it copies the remaining data from mostly "abandoned" erase blocks and flashes (erases) the old erase block. All that and more brings down the write amplification, which measures the average number of sectors actually written for each write to a sector. Intel claims a write amplification of just 1.1 for one of its controllers. Also, wear leveling makes sure that erase blocks are used evenly. Otherwise writing the same few sectors over and over again would burn out the drive in seconds. All in all you can expect to write at least a few hundred times the capacity of the drive, in any order and to any sectors you want, before you need to worry about flash cell wear.
Any wear levelling worth its salt will not do what the grandparent wrote.
Yes, which is where the smarter controllers come in, and where you have the process of "garbage collection". There was a piece a while ago on TRIM not being support on some Apple gear, if I'm not mistaken.
I'm not a complete idiot... Some parts are missing.
RAID in acronym only. The R means redundant, and RAID 0 is not redundant.
Contribute to civilization: ari.aynrand.org/donate
It's only wacky if you bet on it happening next year. But, soon enough.
How soon is soon enough? 5 years? 10 years? And on what physical principle does your argument rely?
When all you have is a hammer, every problem starts to look like a thumb.
The "big win" for solid-state for a lot of applications is shock-resistance.
Most server racks, desktops, and set-top-boxes outside of earthquake zones don't have this requirements but anything mobile does.
Having said that, my ideal laptop would have oodles of storage but the drive would hardly ever need to "spin up" because almost everything I need would fit in the SSD. In "real terms" this would be at least a 128GB SSD plus at least 2TB of less expensive storage.
Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
Indeed. And when reaching larger capacities, it's quite likely that you're dealing with largely sequentially accessed streamed data, ie, video, where you have a maximum needed transfer rate which the HDD is entirely capable of fulfilling which means the SSD gives zero added value for the price premium.
No, it doesn't. It doesn't protect you against losing data in a fire, it doesn't protect you against losing data to malware, and it doesn't protect you against losing data to making a mistake. All changes are automatically propagated across all disks. Backup protects you against losing data.
What RAID 15 does is protects you against losing a day of work because one disk failed - that is, it protects against loss of uptime.
Most manufactures leave any number of gigabytes of flash unmappable for filesystems, that way you can never fill up the drive, even if you fill up the file system. Most pro/enterprise versions of the drive just leave a larger area unmapped.
If you need a 20-30ms initial access time, and then a constant transfer rate of 20-50MB/sec, that makes tape completely useless as it can't fulfill initial access time, and it makes SSD pointless as it overdelivers without added value for everything above that. IE, for bulk data that gets streamed, such as basically any large datasets like video, price per TB is the factor that overshadows anything else.
IOPS is of course hugely important for the average utter crap database written by an intern that devolves into 512byte random access read/write patterns, which seems to be what 'enterprise solution' means these days. But the disasterous consequences of that usually keep the data sets into whatever fits on a comparatively small and cheap SSD as anything beyond basically using processor L1 cache will make the application too slow to use.
SSDs will likely get there in 3-5 years by Moore's law...
Moore's law does not predict prices, it predicts transistor density. Even if you permit yourself to conflate price with density, say there is a factor of 16 to make up, by Moore's predicted semiannual doubling, that is 8 years, not 3-5.
The big fat fly in the ointment is that process costs are rising dramatically and there are yield issues. Plus as you say, magnetic media technology is not standing still. I say, after 8 years have passed, the big price differential will still exist. By that time disks will have moved firmly to an archival/heirarchical role and a typical workstations will only have them as secondary storage, if at all. This will drive up production costs somewhat by eroding the economy of scale, but the world's archiving requirements pull in the other direction, which should put something of a floor under shipped volume. Where will that floor end up? Possibly permanently in the 10X range. Disks just aren't going to die if that happens.
When all you have is a hammer, every problem starts to look like a thumb.
On the other hand, you're also royally boned if you buy a Seagate in the first place. They break unbelievably fast in comparison to any other brand I've tried.
With 3 bits per cell, how long before the data fades?
This is the reliability issue that nobody wants to talk about. I am sure that many others are like myself, with a closet full of old PCs. I like the idea that if I were to pull one out and power it on after having sat unplugged for a span of years, it would still boot (CMOS battery BIOS issues not withstanding) and would still have all of the data I left it with.
SSDs on the other hand won't even guarantee that your data will still be there after *only one year* of being powered off, and as we've dipped below the 34nm process, sometimes SSDs are warranted for even less.
http://cylan.deviantart.com/gallery/
More accurately, recent versions of OSX have their use of TRIM commands limited to the 'apple endorsed' models of SSD, the ones the machine ships with. There's some dispute over the reasons for this. One faction claims it's Apple trying to sabotage upgrades, making it so that if you buy an after-market SSD rather than paying their insane markup performance will become awful. Another faction claims it isn't deliberate sabotage, but rather a lack of interest in testing for unsupported hardware configurations: TRIM can potentially malfunction horrible if the SSD doesn't impliment it in quite the expected way, and Apple has only coded and tested it for their preferred models. By disabling it on third-party hardware they remove the need to test for fifty-odd different devices to make sure it isn't going to corrupt data.
So you would pay $1200 for a hard drive "without hesitation"?
100% Yes for that storage space that ran at Samsung's claimed 1.6GB/s speed...
It would make a huge difference for image management where I'm often loading many 60MB TIFF files in the course of looking over processed images.
Now mind you I'd be backing that up on the cheaper "real" hard drives, but for working with that speed would be fantastic and easily worth the money in terms of saved time and frustration over the life of the drive.
The thing is, that drive will probably be more like $5k which is a much harder amount to take... probably $2k is the edge for my own use.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
Risky Array of Imminent Disaster.
Depends on the application. For a workstation or build box, we configure swap on the SSD.
The point is not that the build box needs to swap, not with 32G or more ram, but that having swap in the mix allows you to make full use of your cpu resources because you can scale the build up to the point where the 'peaks' of the build tend to eat just a tad more ram resources than you have ram for (and thus page), which is fine because the rest of the build winds up being able to better-utilize the ram and cpu that is there. So putting swap on a SSD actually works out quite nicely on a build box.
Similarly, for a workstation, the machine simply does not page enough that one has to worry about paging wearing out the SSD. You put swap on your SSD for another reason entirely... to allow the machine to hold onto huge amounts of data in virtual memory from open applications, and to allow the machine to get rid of idle memory (page it out) to make more memory available for active operations, without you as the user of the workstation noticing when it actually pages something in or out.
A good example of this is when doing mass photo-editing on hundreds of gigabytes of data. If the bulk storage is not a SSD, or perhaps if it is accessed over a network that can cause problems. But if the program caches pictures ahead and behind and 'sees' a large amount of memory is available, having swap on the SSD can improve performance and latency massively.
And, of course, being able to cache HDD or networked data on your SSD is just as important, so it depends how the cache mechanism works in the OS.
So generally speaking, there are actually not very many situations where you WOULDN'T want to put your swap on the SSD. On machines with large ram configurations, the name of the game is to make the most of the resources you have and not so much to overload the machine to the point where it is paging heavily 24x7. On machines with less ram, the name of the game is to reduce latency for the workload, which means allowing the OS to page so available ram can self-tune to the workload.
-Matt
All RAID levels protect against loss of data due to failure of individual drive(s), port(s), or data cable(s).
RAID 0 is not RAID.
RAID is not backup.
It also isn't a backup if the data isn't offsite...which is beyond the scope of any individual hard drive.
Bad User. No biscuit!
My first computer stored data on audio tape! I don't know what their capacity was, but I remember my father borrowing games from work to run through a dual-cassette deck. Some of them were copies of copies, and you had to fiddle with the treble knob to get them to read.
I don't think we're beating that unless someone here is old enough to have used core memory or fluid delay lines.
A 7200 rpm HDD can do 200-400 IOPS or so, semi-random accesses (normal database access patterns). A 15K HDD can do ~400-600 or so. Short-stroking a normal drive also gains you at least 100 IOPS (so, say 300-500 IOPS on a short-stroked 7200 rpm HDD). That's off the top of my head.
A SATA SSD, of course, can do 60000-100000 IOPS or so and a PCI-e SSD can do even more.
-Matt
You can get the flash chips off. Chances are one of them will be dead, but very few SSDs have just one chip - most seem to have eight of them, so you can get seven good ones.
It's not an easy thing to do though. You need a very high level of skill with surface mount desoldering, and an even higher level to reattach them to a test platform to see which ones are duds.
You can have mirrored ram. you can have battery backed-up ram.
Yes you can. Now name me one vendor that actually offers that today....
I'm not a complete idiot... Some parts are missing.
One faction claims it's Apple trying to sabotage upgrades, making it so that if you buy an after-market SSD rather than paying their insane markup performance will become awful. Another faction claims it isn't deliberate sabotage, but rather a lack of interest in testing for unsupported hardware configurations...
Seems like a distinction without a difference.
Second class citizen of the New Gilded Age
Your view of OPS is pretty rudimentary and limited. Are you an intern by chance?
Agent K: A *person* is smart. People are dumb, stupid, panicky animals, and you know it.
Think back, to the day when we used to run everything off floppy drives, and then hard drives showed up. The first hard drive I ever experienced was a 5 Megabyte drum drive. It was HUGE for the time. AND fast. Then spindle drives came out, and were so much faster (and more expensive) AND could fit inside a computer. Then the drives kept getting faster, and more Storage, and then they started to shrink and ....
Now we're running up to limitations of that class of hardware that new classes (SSD vs Magnetic) are starting to impact the Hard Drive space. The change is coming and once the Spindle drive disappears (like floppies), you'll wonder how we ever lived like that.
Agent K: A *person* is smart. People are dumb, stupid, panicky animals, and you know it.
Such statistics are meaningless in my book. Light bulb manufacturers claim their bulbs will last five years or seven years but when you look at the fine print they say that's given under the idea you're turning the light on, leaving it running for 3 hours, and turning it off once per day -- nobody uses light bulbs like that.
But this is the opposite direction as your light bulb example, very few people write 3.2 TB of data every day for 5 years straight, so for most users something else will break before cell writes becomes a real issue.
6TB for $300 is $50 per terabyte, while current pricing is around $400 per terabyte. That's a factor of 8, not 16. I based my math on 18 month doubling, but that's for performance rather than density, so I was admittedly off. Still, that should take you to roughly 3 * 24 = 6 years, not far off my original figures.
In terms of the applicability of Moore's Law to SSD pricing, prices for SSDs have been dropping far faster than Moore's law since the first practical SSDs hit the market. My first consumer SSD was purchased in 2009 at $8750 per TB. Prices today are at about $400 per TB. That's a factor of 22 price drop in roughly five years.
Have you investigated the performance/cost of hybrid versus pure-SSD? I use pure SSD on my laptop and desktop and it's brilliant, but I don't have to cram 2TB+ of SQL Server data stores on my desktop.
Finding God in a Dog
Indeed, I'd forgotten how slow HDD's were until I went to play a game the other day that I didn't often play (and thus was relegated to being installed on my larger 2TB HDD instead of the much faster - but smaller - SSD). The game wasn't too slow to start up, but loading levels resulted in a rather noticeably turtlish progress bar...
Engineers in Taiwan have supposedly solved the cell wearout problem. See http://phys.org/news/2012-12-t...
Like a good neighbor, fsck is there
Yes and no - they simply don't QA every drive that ever existed or will exist, because they didn't ship them and it would be ridiculous to do that anyway. Where the change was, is that they implemented code signing on kernel extensions in order to beef up security a bit, and the side effect is that the ugly binary patch people were applying to the AHCI kext is quite broken. If anyone out there was patching any of the other 200+ kexts that ship with OS X, they have a similar problem; unless they turn off kext signing. Which you are free to do with a simple nvram flag.
Or, it's postulated that you could create your own signing cert, import it into system.keychain, and then re-sign the patched kext; but I'm not aware of anyone doing that. The third option is for some nice SSD manufacturer to make a proper signed driver, but they don't want to spend the time or resources.
Either way, it's a tempest in a teapot. Software vendor is blamed when unsupported hack applied without their knowledge is broken in a major upgrade. In other news, fire is still hot.
Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
Samsung just now announced a 3.2T card? Why are they so late to the party? Fusion-io has had 3.2T cards out for well over a year, and has 6.4T cards out on the market now. Most people won't be buying them anytime soon for home use, but for the person who really feels they need one, they're available. I think the online price to buy through Dell is about $24k right now.
Which is where wear leveling comes in.
Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
Other only that but hard drives have started to become more and more unreliable.
Make that 10-20 years in a hot environment (~60 degrees Celsius). For most intents and purposes flash is a non-volative memory but yes, some applications must treat it as a type of dynamic memory.
Correct me if I'm wrong, but don't SSD's have a point where they put on too many write's per bit?
They do, but it's high enough that your computer and hard drive will be obsolete and replaced well before the point you reach it. Even if you keep your computer for a decade. I'm more concerned that every SSD I've owned has had a serious bug discovered after I bought it.
Crucial M4 - the 5000 hour bug.
Samsung EVO - slowdown bug.
I like SSD's but I have to question the maturity of the technology.
Calling someone a "hater" only means you can not rationally rebut their argument.
Tape drives still serve a purpose. Show me any other removable storage mechanism that can store 2.5TB before compression for $40 and have them still readable after 10 years in a safe.
Spinning rust STILL can't do that.
No, you don't get very good performance from tape, and tape is a general pain in the ass; but if you need to keep something for a long time, and you need it to be reliable, tape is how that's done.
Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
My first computer stored data on audio tape! (...) I don't think we're beating that unless someone here is old enough to have used core memory or fluid delay lines.
Commodore 64 or similar right? Heck, I did that and I don't think that's anything special here on /. it's 80s tech. Now let me get my dad in here so he can tell you all about vacuum tube computers, you kids and your fancy schmanzy transistors. In other words, I think you're solidly beat.
Live today, because you never know what tomorrow brings
And Amazon says the battery in their Kindle ebook reader lasts 4 weeks - fine print: when used 1/2 hour a day. Pssst - that is an actual 15 hours. And they have a graph with the line for their battery dozens of times longer than the competition which has - ready for it - 13 hours battery life.
I guess my car will only need its tank refilled once a year if I only drive 5 miles once a week.
what's funny, is that SSD and tape could completely replace spinning magnetic.
Tape, specifically LTO, is still the long-term backup medium of choice for price, density, and reliability of restore. It's just shit for latency and throughput.
Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
Why do my modpoints always disappear whenever someone comes up with something worthy of them?
If I have been able to see further than others, it is because I bought a pair of binoculars.
They make SD cards with WiFi that will automatically upload pictures you take to your laptop. Not the same, but I wonder if there would be any benefit to setting something like this up in a hard drive as well. If windows crashes, as long as the drive itself has power, it can continue to backup your drive to a NAS or possibly even to a cloud service.
XDInd
They break unbelievably fast in comparison to any other brand I've tried.
;-)
Some people have problems with Seagate. Some have issues with WD. The Seagates in my soon-to-be-replaced colo box both have 50K+ hours on them, and the SMART logs are still clean as a whistle. I have standbys just in case though.
Please stand clear of the doors, por favor mantenganse alejado de las puertas
Are you stupid?
-Matt
They use capacitors to ensure they have enough time to write out the RAM after power down.
Peter predicted that you would "deliberately forget" creation 2000 years ago...
And crucially for this discussion conventional raid* does not protect against silent corruption because it doesn't know which copy is the good copy. I've found SSDs to be far more prone to silent coruption than hard drives were.
* Some filesystems with integrated raid features can protect against this but of course that means you have to use those filesystems which bring their own issues.
note: i'm known as plugwash most places but i screwd up registering that here somehow in the past and now can't register
Drive Writes Per Day is *the* important metric for judging the write load capabilities of a drive. 1 DWPD is perfectly adequate for consumer/desktop use and many fileserver applications but impractical for backing a database, where 5 DWPD is more appropriate. You pay about 50% more for a 5 DWPD drive than a consumer level one, but if it saves you from having to replace the drive 5 times, it's worth it.
Inaccuracies aside, this is an important property of SSD's to keep in mind when procuring them.
set softtabstop=4 shiftwidth=4 expandtab nocp worlddomination
> Would you buy those 15k's new today? What usage pattern would favor 15k's vs ssd's?
Anything that keeps the drive fairly busy writing. Our particular application is backups. Our backup servers write pretty much constantly. SSDs might last a couple of years, they might not.
DVRs for security cameras are another example application that writes pretty much constantly, so again HDDs are a better fit.
On the other hand, SSDs are a much better fit for most laptops, where you want fast boot and physical durability. Each is the right tool for certain applications.
On linux I do just that.
Unfortunately whenever I try doing that on windows it just declares it's created a "temporary swapfile" anyway when I set the swapfile size for all disks to zero. Is there a hidden setting somewhere to really turn swap off?.
note: i'm known as plugwash most places but i screwd up registering that here somehow in the past and now can't register
That's exactly how I use my light bulbs...well maybe not exactly 3 hours - usually more in the winter and less in the summer.
The previous comment seemed pretty clear about the usage though I don't think Samsung was able to actually verify that claim.
So you're saying it does protect against data loss...
A computer with a RAID 1 has a full backup to protect it in the event of a hardware failure. Sure there are lots of ways for that backup to fail, but you can find problems with any backup system.
I think you are spot on. Back in the day of 100GB drives you could easily fill empty the drive several times a year. Even standard def video you'd go through 10GB of data a week easy (~350MB each so about 300 shows compacity). But now go up to 1080p say 1.6GB compressed using normal codex. You are looking at 1-2000 episodes on a disk. So your write/delete cycle is 3-6X less even at 1080p. For the vast majority of people using large storage that is what it is being used for. Yeah I know: "but I have a lot of photos". 1: even at say 10MB a pop do you really have THAT many photos (about 100 a day for 8 years to fill a 3TB drive)? 2: if you do .. and is this personal or business? Yeah a professional might create a lot of content but then storage is a cost of business not just a routine maintenance thing you have to do to organize your personal life.
And 15 years ago I bought a massive 2GB drive for $350.
Computer stuff gets cheaper over time. There's no reason the same won't be true for SSDs. At some point SSDs will be cheap enough that even if HDD are still 1/100th of the price, SSDs will still win because of all their other advantages.
You can get a 32GB SSD for $40 these days. Since 32GB is more than enough for anybody (if you live in the year 2000), there is clearly no reason for anybody to buy anything bigger.
The problem is that while prices come down, the demand for more storage goes up. Today everybody wants to store hundreds of hours of 1080p video. Tomorrow everybody will want to store hundreds of hours of holographic video at 40k pixels cubed. You can never have too much capacity, and spinning disks are likely to stay ahead on that front for a while to come.
Admittedly you said NAS but if NAS really means a box sitting next to the powerful computer that needs that kind of throughput then Thunderbolt will do that for you 20Gbps currently. I suspect the issue is a north bridge south bridge type of thing. Well now the north bridge is really in the CPU for the most part but I/O still lives one step removed from the CPU. To fix that you need to use a pcie card or something for the network interface which means you effectively need the equivalent of a dual graphics card mobo worth of hardware just so that 5% of the time you actually need that 10Gbps throughput on the network interface you have it.
Sadly the ram air intake at the front of the structure does little for aerodynamics. There is too much damping material (soup meat) in the chamber to pass through cleanly to the exhaust port. Additionally, the locomotive openings are often shaped in such a way to create lift at higher velocities, thus reducing overall stability.
Photography is a keen hobby of mine and I have roughly 4Tb of photos dating back to my first DSLR in 03. And yes I do a rather thorough pass of removing photos that are not good, duplicates or otherwise not worth keeping from every shoot I do. Raw files from Canons 22 megapixel 5DMkIII are 40 - 45Mb a pop. Jpegs aren't any use to me.
-------
Drink Coffee - Do Stupid Things Faster And With More Energy!
I don't know what their capacity was...
Well a C90 tape had a 90 minute length and, depending on you computer the data was written at 1200 baud (BBC Model B) to ~1500 baud for a ZX spectrum. Unfortunately there was some overhead so lets say this was 20% (guesstimate). This would give a tape capacity of 90x60x(1200/8)x0.8=648000 bytes or ~633 kB. Some people used to use C120s which would get you an extra 33% but those tapes were thinner and more likely to break or suffer degradation in sound quality which meant you lost your program. With a Spectrum and a C120 you'd might be pushing the dizzying heights of a whole MB on tape.
Late 1980s: I used magnetic core memory attached to the AQS-901 acoustic processing system once fitted to Royal Australian Air Force P3 Orion maritime patrol aircraft. Also had the joy of paper tape and magnetic tape.
Patent litigation: A doctrine of Mutually Assured Destruction... in which everyone seems willing to push the button
In 1987, I bought an 80 MEGAbyte drive for $775 (around $1600 today), thinking how amazing it was that disk drives had broken the $10/MB barrier. When the first 1GB drives came out a few years later, I remember thinking, "Who would trust that much data to a single device? What an amazing single point of failure!" Now there are 128GB MicroSD cards for under $1/GB. Even understanding the technology, the mind boggles.
You got a deal. Around 1989 I sold a 315MB IBM "Winchester" drive to the phone company, for a whopping $10,095 (list price at the time). It slid into a fairly clunky PS/2 Model 80, as I recall.
The subject who is truly loyal to the Chief Magistrate will neither advise nor submit to arbitrary measures (Junius)
Such statistics are meaningless in my book.
Not entirely - it tells you the number of write cycles each cell has: 5*365=1,825 cycles. You then just have to hope that the load leveller knows what it is doing because a single file like a mailbox could easily exceed that in a day if the writes were always to the same location on the disk.
Sorry, no, in IT terms that's not a backup. It's a backup when it's an independent system with a history. Calling a RAID1 a backup to protect against data loss is similar to calling a redundant power supply a backup to protect against power failure. It's kind-of-almost-right in a limited sense, but it misunderstands the problem. It's just hardware redundancy, not a backup.
> Sure, some writes will be happening, but not even close to the magnitude that you'd need to wear out one of these drives.
You mean such as databases doing heavy traffic in badly organized tables, which is one of the most critical commercial uses for SSD.
But that's not the case - there is actually a duplicate of all the data on a second drive. It's only one period of time (the most recent), but it allows you to recover the data in the event of a hard drive failure (and the loss of data on that drive).
Well why don't those engineers review the whole idea of SSD drives. Really, honestly, they seem rather redundant, once the shift is made over to that format, surely storage slots for chips on small daughter boards makes more sense. why bother with a box in a box, when all you need is the slots, just like ram slots to slip in flash storage. Preferable a reasonable number of slots, with the ability to transfer data from one slot to another and say a specific boot slot. Whether you add in additional drive extension identification per slot or just expand the core drive with each additional card installed. On growing failure rates add a new daughter board, mirror the boot card and swap them.
The concept seems outdated for SSD drives when SSD storage becomes dominant, except of course for external drives. Keeping that storage connected directly to the mother board means data transfer speeds will be improved and overall design should be simplified, mother board and computer case, apart of course from the need for optical drives unless of course they get tossed out the same time as hard disk drives.
Chaos - everything, everywhere, everywhen
If they're following industry standards, the typical guarantee on client drives is that your data will be available for 1 year of 60C bake assuming you've already cycled it to the endurance limit.
If your temperature is cooler or you haven't used up all your cycles yet, then retention will be longer than the guarantee.
Enterprise drives trade retention for endurance, thus allowing them to support more write cycles within their warranty periods. The trade-off is that the endurance limit at 60C becomes 3 months instead of 12 months, which for typical enterprise applications is more than enough. Note that when powered up but idle, the drive is performing management of the NAND, so the retention numbers are all assuming that the drive is powered off and in storage.
More data, damnit!
Streaming block writes, a 15k has about the same average write speed as a 6tb 7200 similar cost and 10x the capacity.
Now if your backup or DVR app effectively makes that random writes, sure there is a point but get an app and/or file system that is not broken by design.
No sir I dont like it.
At the largest of the large sizes, hard drives will likely stay behind rotation for another decade when only considering cost.
However, if you don't need terabytes of fast storage, we've already crossed the threshold where SSDs are cheaper.
Smallest hard drives you can buy these days are $60 new and store 500GB. That same $60 gets you a 128GB SSD from a "Tier 1" manufacturer (Samsung, Intel, Micron/Crucial, Sandisk, Toshiba)
For just about any storage application that fits in ~100GB or less, SSDs are both cheaper and more reliable TODAY than rotating drives.
That 100GB crossover threshold with the cheapest rotating drives will double every year or so, since today's rotating drive prices are almost completely based on the cost of the electronics and a single head/single platter mechanical system. You can't make a rotation drive significantly cheaper than today, but with each generation of SSD they can halve the number of NAND packages, shrink the PCBA, build controllers with fewer channels, etc.
More data, damnit!
That was true over 3Gbit/s SATA perhaps, but hasn't been true for a while now.
A single PCIe SSD in a 2.5" form factor can move sequential data at 2.5GB/s or better (Gen3 x4). Modern rotating drives cap out somewhere around 200MB/s if I remember correctly, even on 6Gbit/s SATA or SAS links. That means that for sequential IO, you need 12 rotating drives to match the performance of one SSD. For random reads (700K/s on the latest samsung) you need 5800 rotating drives (assuming 120 IOPS from the rotating disk). Note that would require a queue depth of almost 700,000 which is impossible from an application standpoint. In reality, your pool of 5800 rotating drives will be MUCH slower.
More data, damnit!
Erases can fail, but that's typically a gross failure in the peripheral circuitry and not a cell-level/array-level problem. It's no different than you being unable to erase your data if you have a mechanical failure on a rotation drive.
Your most likely "leakage" case is with a grown defect or a change in the flash translation layer, however, the specs are written so those old locations must be erased by a secure erase command. I know that based on NAND physics, if you do that erase, the data is gone and never coming back. IMO, there really aren't enough electrons in a charge well to reliably encode "additional" information about the prior state of a bit following an erase.
An NSA hack is always possible where they install rogue firmware on the drive that doesn't actually secure erase properly, but that kind of argument/speculation is outside the scope of my answer.
More data, damnit!
New shingled harddrives have some very complex data-keeping going on, about as bad as SSDs. Pointers everywhere, keeping track of where the LBA actually goes on the disc.
And then we can slowly pull the cards out of the slots while the computer sings Bicycle Built For Two in ever slower verses.
-- I ignore anonymous replies to my comments and postings.
You might as well ask the same question about a hard drive. If you power down a hard drive and put it on a shelf for a year, there is a better than even change that it will be dead when you try to power it up again, and an even higher chance that it will die within a few days.
A powered-down SSD that has been written once should be able to retain data for ~10 years or so. Longer if kept in a cool place. As wear builds up, the retention time drops. You can look up the flash chip specs to get a more precise answer. A powered-up SSD should be able to retain data almost indefinitely as the self check will relocate failing sectors as they lose charge. However, in practical terms, it also depends on how the drive firmware is stored. The drive will die when the firmware is no longer readable. But that is true for hard drives as well.
-Matt
Modern wear levelers will even move blocks that aren't being written to, in order to maintain evenness. Older ones only allowed wear leveling to happen with "free" blocks, but now all blocks can be moved.
I'm not the most tech savvy guy in the world, so I'm certainly willing to be corrected, but my major problem with SSD's is this: when they fail, they do so without warning, and in a way that makes even partial data recovery impossible for an average user. On the other hand, in the decades during which I've used HDD's, I've never once had one fail without giving me some kind of warning. The dying drive has either overheated, or started showing read/write errors, or made distressing noises (kind of like that death speech all the soon-to-be-croaked best buddies get to make in action movies). And I've even been able to recover at least some data from drives that were seriously screwed. The only failed SDD I ever dealt with simply didn't report on boot, and I never heard anything from it again. So to me, a big SSD just offers a better chance to lose everything all at once, minus whatever was saved in the last backup.
And then there's cost, of course. I just bought a 2TB HDD for eighty bucks. I know I won't be seeing any SDD's available at that capacity/price for a long, long time.
Right now, my ideal computer would have a mid-size SSD for the operating system and installed programs, and a big, fast HDD for most storage needs. And, of course, my external backup drive would use old-fashioned platters. I'll leave huge, relatively expensive SDD's for those whose need for speed is much greater than mine.
I've calculated my velocity with such exquisite precision that I have no idea where I am.
Gimme gimme gimme.
Got Geometrodynamics? Awe, too hard to figure out? Too bad.
HDDs are not as recoverable as you seem to think. I have several bricked drives to show for it. Plus there is a trade-off in that your HDD's chance of failure goes up dramatically over time no matter how little or how much you use it. Even keeping it on a shelf won't make it last longer. SSD failure mechanics are very different beasts. If your SSD is barely worn after 3 years of operation (and most will be), the failure rate will not be appreciably higher than when it was new. The chance of multi-bit failures eventually overcoming the automatic SCAN/relocation (in SMART) will increase once appreciable wear occurs, but the wear is write-based and not time-based and for most SSD users that means reliability will be maintainable far longer than the 3 years one can normally depend on a HDD for (assuming it isn't one of those 5% of HDDs which fails every year anyway).
And, again... You don't make backups? Depending on the recoverability of your hard drive virtually guarantees that you will lose all your data one day.
-Matt
A RAID can be lost or corrupted, or someone can overwrite or delete a file.
And tapes can be lost or corrupted, or someone can burn the building down.
This is an old argument, and every time it gets revisited RAID starts to look better. Overwriting / deletion might have been a concern prior to modern filesystems which incorporate easy and cheap snapshotting, but nowdays that part of the argument just doesn't fly. Corruption is still a concern but, again, that's a risk you take with any backup solution too.
There's no such thing as a guaranteed backup. If you're very rich and very paranoid, you could certainly rig up a "backup solution" that involves copying your data every 5 minutes to 50 different offsite locations in 50 different countries, plus having some cheap third-world-labour transcribe all the zeros and ones to a paper copy for storage in an underground vault. And even that's not 100% because a really big asteroid will result in unrecoverable corruption. In the end it all comes down to how much you're willing to spend and what level of risk you're willing to accept. For most of us who aren't running IT departments that equation comes down to something like "ZFS RAIDZ2".
Why use unreliable 15k drives for backups when you can just use 7k2 dives for the same price and sequential speed but far more room?
Sequential speed is not just spindle speed, it's also a result of density. Sequential speed increases with the square root of the density increase. A 2x as big disc with the same spindle speed will be 1.4 times as fast sequentially. A 4x as big disc will have twice the sequential speed.
Thus a 4x as big 7k2 disk will have almost the same sequential speed as a 15k drive, but with 4 times the room. Of course you should still stick them in a suitable RAID.
Well, I might have a way, but it only works on a semi spherical planet in a vacuum.
But it doesn't allow you to recover data in the much more common event that someone mistakenly erased it. As you'll restore about nine files due to mistakes for about every one you'll restore due to disks failing, that's what backup is supposed to protect you from.
Also, RAID ignores two other major failure modes, and that's faulty hardware/bus, and filesystem software bugs that hurts/destroys your entire filesystem. RAID won't help you from that either. In fact, since such bugs are relatively more common in professional RAID controllers, you're slightly more at risk from those when you run RAID, than without. (As any pro and they'll tell you a story about when the entire RAID-array failed horribly.)
So, no. While RAID can be an important part of any availability strategy, it's not "backup" for any useful definition of that word.
Stefan Axelsson
That has nothing to do with using RAID or not. Even normal backups wouldn't survive that, unless they are in a fireproof container or stored off-site.
So what? A meteor could crash on your computer too. RAID still offers extra protection against loss of data.
You must be joking. Mechanical hard disks do not have wear leveling. They only have some extra space to remap some individual failed sectors. The media of a mechanical HDD does not really wear out anyway.
The change is coming and once the Spindle drive disappears (like floppies), you'll wonder how we ever lived like that.
I already wonder that when I see my system sitting 99% time in iowait if a modern CPU is paired with any mechanical HDD. It's an almost unacceptable bottleneck.
Now what about storage durability? With 3 bits per cell, how long before the data fades?
I was under the impression that the controller would handle this. Cells are typically marked as dead once their thresholds are such that you can't guarantee that they'll hold their contents for a year (there was an interesting paper at EuroSys this year about extending the lifespan by using these cells for short-lived data and exposing that functionality to the OS). If a cell is getting close to the time when the data has been unmodified for long enough that its integrity is in danger, then the controller will use the same mechanism as wear levelling to read it and write it back (either in the same place or somewhere else). Most of the time, this will happen as part of normal wear levelling, as unmodified data are moved around to sit on cells that have been rewritten a few times and spread the wear onto some of the cells that were only written once.
I am TheRaven on Soylent News
When I started university the central ICL 1906 mainframe had 384 K words of core, and a year or two later we upgraded to 512KW (24 bit each) of "solid state" memory with the unbelievable access time of 0.3 microseconds. The paging devices were drums because of their much lower latencies compared to disk drives - we had quite a few drums with a couple MB. No mercury delay lines but the VDU display memory was coax delay lines. At least we didn't have to submit our programs on punch cards like the ME and EE peons - we were CS so we had *online* access!
thegodmovie.com - watch it
Moore's law has been dead for years. It applied to Flash only in the beginning and the cells have reached physical limits, that's why it's only about cramming more cells into a package today.
thegodmovie.com - watch it
There's a significant effect though: The market for lower capacity HDDs has disappeared, and with it much of the volume so the HDD manufacturers have to make their margins at the higher end. 1TB HDDs are an endangered species now because you can get a 120GB SSD for the same price and many consumers don't need 1TB while they appreciate the performance of an SSD.
thegodmovie.com - watch it
The consumer-grade SSD in my laptop can happily handle 2-300MB/s of sustained writes (and, simultaneously, 200MB/s of sustained reads). If you're doing linear writes, then you're the optimal workload for wear levelling. You'll be hard pressed to find a drive that isn't guaranteed for 5 years of writes at the maximum throughput the drive can handle.
Although, as other posters have pointed out, you'll get better sequential write speed and reliability from a RAID array of slower disks.
I am TheRaven on Soylent News
They use capacitors to ensure they have enough time to write out the RAM after power down.
Only problem is, it doesn't work very well in practice:
http://www.extremetech.com/computing/169124-the-mysteriously-disappearing-drive-are-power-outages-killing-your-ssds
Not really, an easy way to falsify your account is to run an old motherboard, hard drive or pretty anything that's not been used in ten to fifteen years : it works fine. The firmware is usually stored in flash (instead of PROM or EEPROM as was done in the 80s and early 90s) which is also why all it takes is a DOS program to update firmware instead of removing chips and putting them in a programmer.
1.8" is dying, but you have mSATA and M2 interfaces instead.
Thunderbolt is external PCIe, and is thus nice for overpriced laptops and trash can shaped workstations. Else it is a lot cheaper to use your internal PCIe. PCIe 2.0 4x will do the same job as Thunderbolt 2.0 - and so you can either use a PCIe 2.0 4x card for your SSD (or a 3.0 one to yet double the bandwith, seems the 3.2TB Samsung SSD uses that) or for a 10Gb NIC.
Real issue with 10Gb ethernet is the cost and then the power use, not anywhere near a graphics card but around the 10W mark which is significant. A high end motherboard was just announced (2011-3 socket) which has a dual 10Gb NIC built-in, which uses 14 watts.
Indeed, complicated storage schemes tied to a hardware controller that's likely to fail (a computer in your computer), what could go wrong.
Better to use software RAID 1, or even rsync a drive to another.
In the context of home storage anyway, I would say RAID 1 (and checking your RAM yearly with memtest86+) is better than nothing at all, which is the "backup strategy" adopted by 90%+ people. Better yet would be RAID 1 + off-site backup. Note to self : have to make a list of files and folders in my music collection. That would be a useful "back up" already.
If you actually use a kindle, you will realize that the 4 week claim is quite true. I do not know the fine print etc but I have been using one for several years and it still gives me a nice 3-4 weeks battery. I read about 2-3 hours daily.
Wealth is the gift that keeps on giving.
And tapes can be lost or corrupted, or someone can burn the building down.
This is an old argument, and every time it gets revisited RAID starts to look better.
This isn't a competition. I'm not saying, "Screw RAID! It's a terrible backup." It's just not a backup. I'm not going to fight with you over this. Go ahead and use RAID as a backup. Maybe you'll be lucky and you'll never need to learn your lesson.
I'm using a 12-year-old Seagate fine. Yay, really a drive that old still is big enough to run the latest OS and it is nicely quiet compared to a Maxtor or worst, Quantum.
Seagate had its shit series that's not good for anything, 7200.11.
but it allows you to recover the data in the event of a hard drive failure (and the loss of data on that drive).
Well, it allows you to recover data in the event of a hardware failure specifically on one of the hard drives, nowhere else, in such a way that doesn't cause data corruption first. In much the same way that if you have a redundant power supply, it will protect you against the specific event of hardware failure where one of your power supplies fails without there being a problem with your power source or damage to any of the other internal components.
That is to say, it's hardware redundancy. Nothing more. Of the events that lead to data loss or power failure, hardware redundancy does protect you against the case where the problem is limited to hardware failure of one of the redundant parts, and everything else works properly. A "backup" however should be a more generalized strategy for protecting against a total loss of the service, i.e. power goes out or data is lost.
If you're worried about that it's easy to not fill it up, make partitions so you leave out 10 to 20 GB that will never get used at the end of the (logical) drive, in addition to the overprovisioning that is already there.
Perhaps the OS needs an option to artificially slow down the swapping. Or just get over it. If you're too worried about the various caches, temp files etc. eating your precious I/O reserve then your SSD will be wasted because you're afraid to use it.
On a desktop you will have gigabytes that come from dailywtf-worthy javascript from the internet, pixmaps and buffers that came from rendered html and jpeg files etc., that sort of crap. If you leave that shit running for days (tab hoarding, even mild) and the OS puts some old unused shit into swap so you get more free memory and disk caching, that seems a win.
Consuming all RAM with only an OS and browser is also all too easily doable. So I would say you have it more easy on a server (unless maybe you run garbage software there which I'm sure there's no shortage of)
So a 3TB drive is enough to store 15 pics a day for that 11 years a 50MB a piece. I assume back in 2003 you weren't taking 50MB pics but going forward it might grow I suppose. But still say you take 150 pics a day and then delete them all once you fill up the bad boy: it is still over a year between cycles. With at least 300 write cycles on an SSD you'll be fine: in 5 years you'll be picking up a replacement 10TB SSD for $200 from Best Buy.
Yeah a 10W nic is drawing "a lot" of power but so is the disks/switches to get you data at that speed. SSDs are low powered but you still are looking at using a few of them in parallel to feed you that quick and presumably the NAS device also must have a nic on its end. Not a huge issue other than heat though since I don't think 10Gbps makes sense on a laptop much yet, well I guess it does because at that speed you don't need much local storage, but if you were pulling stuff into local storage you'd fill up laptop class drives in about 10min of use.
> You'll be hard pressed to find a drive that isn't guaranteed for 5 years of writes at the maximum throughput the drive can handle.
Hah! I wish. The most popular line of SSDs is the Samsung 840 series. In commercial usage, the 840 Pro is warranted for 73 TB written. That's 3 DAYS at maximum throughput.
The 850 series is warranted for twice as much - 150 TB, or one week at maximum throughput.
The biggest problem with the Kindle is it's very easy to forget you need to charge it "sometimes".
If there was ever a device that would benefit for long range wireless charging, even at just a trickle, it'd be the Kindle.
Indexing? Think again. The OS does alot of writing to the HD. So depending on how you have you system configured, it could minimal writes or alot.
But HD failure is huge with SSD. Once it dies, good bye data. Regular HD, you can recover data. Long term data storage is not for SSD.
What, speed? That's the only advantage.
They have 5 today as well. Western Digital Black drives have 5 year warranties.
LMOL yeah sluggish. Considering how long SSD capacity has been stuck at 320 MB, I would not say they have been progressing fast. Couple problems with HD, the OS not being able to see sizes over 2 TB. EFI was a recent development and not all systems run them. Then there were density issues. Those two seem to be overcome.
Wake me when I can get a 1TB SSD for $70.
Because my computer will boot up 1 second faster? Wow.....performance gains are to be had with memory, size and speed.
ECC memory triggers an even on error. There are counters to see how many memory errors you've had since boot. After the first multi-day memtest burn-in, you should be good just monitoring the counters.
5 years for current-gen TLC
That's good Samsung's new cheap SSD drives come with a 10 year warranty.
EFI was deprecated 10 years ago for UEFI. Unless you're talking about 10year+ old machines, they support 2TB+ boot drives. I don't mean machines sold in the past 10 years, I mean the actual age of the tech. Some bargain bin computers are several years old at the time of purchase.
I think the point is that the parent has obviously dealt with IT people that think RAID = backup. I have as well. It is painful.
Also, saying "RAID protects against *some* data loss scenarios" isn't accurate. It protects against one, and only one, data loss scenario: drive failure.
ALL other data loss scenarios are immune to RAID.
One =! some.
Grammer Nazis - I mod you "troll" unless you actually add something on-topic. Yes, I know I have mispellings in my sig.
I loved that line of enterprise 500gb drives they made once that all failed on the same day across multiple data centers... fun times.
A 4x increase in resolution is much less than a 4x increase in storage requirements. Unless we switch to holographic recordings, storage requirements for media is being outpaced by storage growth. Videos have gone from 100MB/hour to 1GB/hour(1080p) in the past 15 years but HDs have gone from 10GB to 8TB, and still cheaper and faster. For only several thousand dollars, I could build a RAID6 setup that could hold every movie I've ever watched in BlueRay quality. 15 years ago, I couldn't store more than a few days of binge watching of blurry videos for the same price.
That's good they learned to stack silicon vertically and dropped my SSD's price by 50% in the past 8 months. I paid about $120 for a 120GB during tax season earlier this year, and now the 240GB model is going for $120. 50% price drop in less than a year is pretty good. The new 120GB models are $115, but they come with a 10 year warranty and are much faster and lower power.
For most of us who aren't running IT departments that equation comes down to something like "ZFS RAIDZ2".
For those who aren't running IT departments, rsnapshot daily, weekly, monthly and yearly is easier and cheaper. Or any versioning backup.
Easier because most such people don't work on BSD / Solaris - so setting up a NAS is the extra step making ZFS difficult.
Cheaper because a lot of data will turn out to be not worth backup - rsnapshot can more easily exclude files that don't look like they need backup. So in a typical scenario, 10 timed versions of 1 TB can stay in 1 TB. Lots of nested mounts can exclude ZFS from RAIDZing too, but it is a hassle far bigger than simple regexp.
Bingo Dictionary - Pragmatist, n. A myopic idealist.
Intel 320 8mb bug too but that was firmware. I've not had any issues at all with my crucial or Sammy Evo 250gb though...
Yes, I'm not saying that RAID has no place. I use software based mirroring for my "big" drive that has stuff that's annoying to lose, but not critical. And then "offsite" (well in my basement...) backup for the stuff that has to be there in addition to mirroring.
But making a list of the filenames, that's a novel idea that sounds about right. Have to do that. (And remember to include it in the offsite backup. :-) )
Stefan Axelsson
A powered-down SSD that has been written once should be able to retain data for ~10 years or so. Longer if kept in a cool place.
Nope. Most MLC SSDs will lose their data in about a year and the TLC SSDs in about 6 months of being powered off. (Don't confuse older flash media which was probably SLC with newer MLC/TLC media. Or which had larger feature sizes.)
As the size of the feature that stores your bits shrinks, so does the archival lifetime before something bad happens to one or more of the bits. That holds true for everything from tape, to hard drives, to CDs to flash drives.
Wolde you bothe eate your cake, and have your cake?
The write count is not a problem : write count exceeded, change drive, copy data, done. It's not like the drive will explode once the threshold is exceeded. And chances are that you will never encounter this problem unless you have a *very* intensive usage.
Perhaps more problematic is fading. Flash cells lose charge over time, and the more you write, the more "leaky" the cells become. That's why we say that the number of writes is limited, because past a certain point, the manufacturer cannot guarantee that the data will stay for more than X amount of time. FYI, X = 1 year for the Intel 520. Note that that's 1 year powered off. If the drive is on, the cells will be periodically refreshed, making the actual duration much longer. ... but optimized for long-term storage at the expense of other things like storage density.
But SSDs are not the only kind of storage medium that lose data over time. Magnetic storage is slowly erased by the earth magnetic field, reflective layers on CDs oxydize, etc... It means that the only way to ensure long-term data integrity is to actively maintain backups, or, if it not possible, use special archival grade storage media. Archival media can be of any kind : (EE)PROM, magnetic, optical, paper,
2TB hard drives that could be bought for $55-70 back then weren't WD Black drives. Even cheap regular and green drives had a 5 years warranty a few years ago.
In my opinion raids are good overall, I use them at work and at home. They offer performance and data integrity, in all configuration except raid 0. Most people i meet that bash on raids on raids either had a bad experience with early raid controllers or don't understand what a raid has to offer let alone how to configure one properly. Modern raid controllers from reputable manufactures are extremely reliable, fast and can offer large disk space with extreme I/O rates. I regularly parse network/application data in to databases that grow in excess of 500GB for a single day of testing there is no single Drive/SSD solution that will work for us. This problem is also compounded by the fact that we may need as many as 20-30 of these databases on-line for the duration of the event. The only solution out there is large disk arrays that allow us to read/write large volumes of data quickly and handle it till we are done. After 8 years of working with raids mainly 5/6/10/50/60 i have yet to lose a single MB of data to do raid failure. This is not to say I haven't had hardware failures but if your mindful of your equipment and keep it cooled and maintained properly most disk array solutions will outlast its usefulness by a good while.
I spent 10 years in Silicon Valley doing PR for flash and other nonvolatile memories. In 1997, we thought we were having Hugo Gernsback visions of the future when we were talking about flash eventually hitting $1/MB. Seems kind of silly now.
A modern integrated Intel 1Gb NIC's silicon with all of the offload features a desktop can use can cost as little as a few pennies extra. 10Gb NICs require their own chips, which are quite large, and need heatsinks. A basic 1Gb switch will consume nearly 1/100th the power of a 10Gb switch when idle. At load, it's closer to 1/10th, but that's not too bad when you realize 10gb is 10x faster, but most home users are idle 24/7. Even when you're watching Netflix or downloading torrents, your switch is nearly idle. My 28 port 1gb managed switch idles around 5 watts at the wall. A basic 10gb switch idles around 200-300. Now there are fans involved, small fans, that spin really fast, and require ear protection to be near. In a house environment, with dust, you'll need to service these switches and their fans.
ZFS defaults to at least 2 copies of all metadata, but the super blocks are special, so they're replicated an additional time every certain offset of the LBA, on top of keeping the last 256 versions of the super block. You may have 4+ copies of each of those 256 super blocks on a modern multi-terabyte HD by default. A new super block is created every 10 seconds by default, assuming there is any data to be written, this is the "transaction group". It's going to take a lot more than 1 block going bad for catastrophic failure.
I'm seeing this on the Internet. I saw slightly different numbers from many sites, but they were all nearly the same +- some percentage.
7,200 rpm SATA ~75-100 IOPS SATA 3 Gb/s
10,000 rpm SATA ~125-150 IOPS SATA 3 Gb/s
10,000 rpm SAS ~140 IOPS SAS
15,000 rpm SAS ~175-210 IOPS SAS
One series is made in Indonesia, the next in Malaysia, the next in Mexico, parts from different subcontractors etc. They all bear the name Seagate but they're not all equal.
But big sequentually accessed files like video or music are perfect for a hardddisc, It's random access & thousands of little files where SSD's shine because of zero seektimes.
This is true for a lot of stuff. Buying a Fender electric guitar? You can get one made in the U.S., or Mexico, or Japan, or China, all priced/featured differently and of varying quality.
Please stand clear of the doors, por favor mantenganse alejado de las puertas
Because, as we all know, performance comes in only one flavour.
This is an even sneakier version of what Daniel Dennett calls "rathering". This is where you write "The proponents of A would say that A resolves this issue. As we can see, A does not solve the problem, so rather B." The trick here is that no-one ever said the issue was a dichotomy between A and B. It's been implied by a rhetorical device that few readers even notice. Apparently Stephen J. Gould used this technique a fair amount. This surprised me. He was a pretty solid author for the most part.
Do you really think that SSD is the best storage option for Google Earth's highest resolution imagery of the Nunavut territory? I guess your philosophy is that if the data isn't in high enough demand to justify SSD performance levels, there's no point keeping the data online in the first place.
Then there's a few hundred people who charter expensive hunting trips in the Canadian north and afterwards they go to Google Earth to review where they've been and Google Earth says "Imagery 404: not enough demand to make it cost effective to host the data on SSD".
If it's just a few hundred people, so who gives a shit?
But big sequentually accessed files like video or music are perfect for a hardddisc, It's random access & thousands of little files where SSD's shine because of zero seektimes.
Sure. However, there seems to be some kind of argument that SSDs will completely replace hard drives for consumer use. That doesn't seem likely to me. SSDs are great for many things, maybe even for most things, but there are many common use cases where they just aren't adequate. Their cost may very well come down, but so do hard drive costs.
...today's rotating drive prices are almost completely based on the cost of the electronics and a single head/single platter mechanical system. You can't make a rotation drive significantly cheaper than today, but with each generation of SSD they can halve the number of NAND packages, shrink the PCBA, build controllers with fewer channels, etc...
And magnetic media continues to improve its linear and areal density at a rate not much different from improvements in flash density, largely maintaining its cost differential vs flash. As I alluded to above, the net effect is to push magnetic media out of workstations (and personal devices), and also out of high value data center applications. However, the mass storage, archival, and heirarchical sectors that remain are huge and growing rapidly with no sign of slowing. This means that rotating media will not be going the way of punch cards for many years to come.
When all you have is a hammer, every problem starts to look like a thumb.
6TB for $300 is $50 per terabyte, while current pricing is around $400 per terabyte. That's a factor of 8, not 16. I based my math on 18 month doubling...
I stand corrected, the price differential has narrowed recently, however a factor of eight (8.4, more accurately) is still plenty big enough to keep flash out of many mass storage applications. This relationship whill gradually change, but it will take a lot more than 6 years to erode rotating media's share of mass storage to something similar to tape, say, where economies of scale start to be eroded. Just pulling a number out of my ass, I think we are looking at 15-20 years before rotating disks start looking like the tape market. After all, you're still going to want a hard disk to store your bootleg videos.
When all you have is a hammer, every problem starts to look like a thumb.
The question is, will we get to a point where it won't matter? There will be a point where solid state drives will be cheap enough that you'll be able to get enough storage at a low enough price that people generally won't care. For example, if $50 got you a 500GB SSD or a 2TB hard disk, how many people would pick the hard disk? Even then, those who want more storage might just pony up the extra cash.
Supplementary storage will be the last to go, so perhaps your timeline is more accurate there, but I think we'll see hard disks almost completely disappear from new computers long before then. It's already starting to happen: many notebooks use SSDs now, from the really cheap Chromebooks to the really expensive retina macbooks.
The question is, will we get to a point where it won't matter? There will be a point where solid state drives will be cheap enough that you'll be able to get enough storage at a low enough price that people generally won't care. For example, if $50 got you a 500GB SSD or a 2TB hard disk, how many people would pick the hard disk? Even then, those who want more storage might just pony up the extra cash.
You seem to be limiting your thinking to consumer storage. Think data center, where even a small cost per byte saving can drive the engineering decision. With the quasi-panic migration to the cloud currently underway, data centers will make up an increasing slice of the pie, including offloading a lot of consumer computing. In the cloud, the long access time of spinning media usually does not matter, it is small compared to other lags. In the cloud, most data storage is bulk data, simply because there is more bulk data. It is going to be many years before any operator that needs to be competitive will be able to afford a wholesale migration to flash.
When all you have is a hammer, every problem starts to look like a thumb.
This amused me. Thank you kindly. ;)
I do not want your cheap brainburning drugs. They are useless for work. And I am a working man today.
If price/megabyte is all you need, I'd suggest to remove your hard-drive and replace it with tape. Yes, the seek times are a bit long, but then again, you didn't care about your time, right. Just price/megabyte.
A single Sata 6Gbps SSD that reads at 500 MB/s is reasonably close to the 10Gbps ethernet bandwith, and M2 PCIe 2x matches it.
I can agree it's not critical at all to have such network bandwith. It's doable right now essentially as a luxury (Apple sells an external Thunderbolt 10Gbe NIC at around $1K as an option) unless you're working with lots of content creation and convince yourself it's damn nice.
The real killer issue is cell size. Smaller cells have shorter lifespans and are slower
For this reason Samsung backed off from 20nm substantially when they perfected 3d technology. The result is 10 year warranties on the 850pro family and there's no reason to disbelieve them on it.
You might as well say "64GB RAM in a consumer desktop? Yawn, my server has 1TB".
It is a consumer SSD, on a smallish card (single slot, single PCB) meaning you'll be able to order it from newegg et al. at one tenth the cost.
Yeah I'm in the it's nice camp. I mostly consume data. I develop software but that is a pittance of the bits I touch a day. Even counting for test builds and what not it is probably less than 1GB a day. It is really people generating uncompressed video that need this kind of throughput on a local machine. But those I'd lump into: 1) professionals so that is the cost of business or 2) amateurs so that is the cost of your expensive hobby (same like gamers piss money away on multi-GPU, liquid cooled etc systems, or hobbyist golfers on memberships etc). It is the fact that the average person doesn't need it that makes it a luxury and gives little incentive for your $1200 Dell to come with it. The vast majority of people are using a NAS for household backups and to push a compressed video to the tv or whatever. 1Gbps is more than enough for this.
You have more data on that single track rotating at 7200rpm, or more independent heads to access data without a seek. I wouldn't call this "speed hitting a brick wall". Furthermore, you don't have to choose one or another. A hybrid drive with 1TB SSD and 10TB spinning disk would be extremely useful, and much cheaper than an all SSD solution would be in foreseeable future.
That's BS. I recently found many smaller USB keys and SD cards that haven't been used for at least 5 years, and they were all fine.
I've got better things to do tonight than die.
Were they as good as the IBM Deathstar?
Ok, so you probably aren't going to read this 3 days later, but just to note that most RAW file formats are highly compressible. So if your archiving decade-old RAW photos there's little reason not to pass them through gzip or bzip2 or lzma and cut their size in half or more.
In my opinion raids are good overall... Most people i meet that bash on raids on raids either had a bad experience with early raid controllers or don't understand what a raid has to offer let alone how to configure one properly.
Wait... who doesn't think that RAIDs are good? RAIDs are... well it's not ever really an issue of good or bad. They're a pretty fundamental part of modern computing, particularly when dealing with servers. But that's not the point. The point is, they're not really a backup. That's not bashing RAIDs any more than it's a complaint against hard drives to say that a hard drive isn't a video card.
The Fusion-io cards are also single slot, single PCB. The 3.2T card is even a low-profile card. So it's more like saying "hey, there's a 32GB DIMM out now!" While true, it's been true for quite some time. They're expensive, sure, but the home user who wants one can still stuff a system full of them.
Also TFA describes not a drive, but a PCI card. I can readily source a 1U chassis with 10 SFF disk bays, but not with 10 PCI slots. And PCI cards aren't nearly as hot-swappable. So more needs to happen before a product like this fills more than a niche.