SSDs have a 'bleak' future, researchers say

[Night Hawk]

For me, it is a moot question, because until the prices of SSDs falls drastically, I shall never own one...large or small.

For me the option isn't even listed but already put to use! Instead of memory driven internal drives I frequently use the larger mechanical internally which provides work space as well as OS and storage/backup.

For external the memory driven usb flash drives provide some other options besides being pocket sized. A 16gb or 32gb model allows me to run the "other guy's OS" or transfer data between machines and carry in a belt holder at the same time for far less while still seeing some interesting prices on the even larger flash drives no doubt.

I did have the x64 7 installed on one of two Sata 3 drives but ended using them for storage/backup while using a pair of Sata IIs for OS purposes for far less on all 4 HDs over what you would pay for just one larger capacity SSD. Yet I'm not suffering performance wise either.

 

[unifex]

I too use the option #1. I never thought of it in terms of price per GB, but for my 60GB SSD I paid significantly less money than for my previous 1 TB purchase. I put my OS on the SSD and the system is noticeably faster.

Now, the main problem with the SSD is not shrinking but limited lifetime. You can't simply overwrite the SSD over and over again. This is the reason why I put certain folders (not simply storage) on the usual HDD.

Shrinking however has a very natural limit that nobody is going to the break - a typical distance between atoms in a solid is 0.5 nm. So even if you figure out a way to make memory bits out of single atoms, you still would not be able to shrink your devices beyond that limit. 6.5 nm is not too far from it, so I don't see the point of that objection.

 

[JimLewandowski]

I just want a mechnical drive with huge amounts of cache storage such that writing huge files results in instant return to the app and destage at the device's relative leisure. I've never played with, for example, getting a large, let's say 30 MB file in word and hitting "save" and seeing how long it takes for the controller to get the data (i.e. cache policy allows asynchronous writing). Then make the file larger than disk controller cache storage and see how much mechanical delay is now introduced.

 

[Night Hawk]

What I have to look at here is not something like gaining an edge as far as gaming performance having the OS running a little faster due to the two different types of drives but the need for drives that will handle loads like video capturing moving volumes of data and for a second identical drive the testing portion of multiple OS installs being likely for varying setups.

Wipe a test drive over and over again depending on need is what that would amount to as well as having plenry of drive space available for C is another factor to weigh in here. For testing the W8 Preview which will be here soon enough I would also look at the backup features which would include full system images being created from and restored to the test drive itself.

Drive wipe after drive wipe is what you wouldn't want for an SSD however from the information is pointing at. Durability over speed gain is what that comes down to. For someone else it would be an OS drive for performance with a mechanical for storage. Here I would compare that to a work horse drive where the mechanical drive will take the pounding over a lengthy period of time as opposed to an SSD failing from being overworked?

 

[profdlp]

"it's not going to be viable to go past 6.5nm ... 2024 is the end."

hmm, there are a whole 12 years of technological development to go until then.

in three years time, someone will invent an ssd super-duperlizer, then in five years that will get superseded by a mega-wegalizer etc etc - probably.

would you have listened to hard drive manufacturers a dozen years ago if they had said "well, 5 gigs looks like the limit - we'll never get past that.", or ram manufacturers saying "we're never going to be able to make a 64 meg ram stick."

Exactly.

At my first computer job I remember my boss (who was a really sharp guy) saying that 4GB was the absolute top due to the limits of FAT drive structure.

Then some bright lads created a drive structure with two FAT tables.

 

[lehnerus2000]

Stock Market Idiots

They may have been 4-5 years ago, the first batches were horrible with reliability, but today's SSDs have a metric crap ton of tricks and over commitment of ram to handle these very well known and understood problems. NO ONE knows about these problems more than the people manufacturing these drives.

That's probably true.

However given the current economic model (i.e. your assets are worthless, only your share price matters) there's no way that they could/would mention it, as the stock market idiots would instantly panic and destroy the company's share price.

 

[GeneO]

I think SSD will be supplanted by some new faster technology before HDD ever catch up in any way. So in that sense they are doomed.

 

[carwiz]

And who said the drives have to get smaller? The drives are small enough now that manufacturers could revert back to the 3.5" format and easily quadruple the capacity with current technology. And think about what these drives could do if they had integrated RAID-type logic or multi-channel processing like RAM. The mechanical drives will never catch up in speed or rate of capacity increase. They have magnetic recording limitations similar to the SSD's electro-capacitive storage limits. 1TB HDDs were hitting the market about the same time as 20GB SSDs (About 5 years ago). Comparing rate of capacity increases in SSDs, how many 25TB HDDs do you see on the market?

There has always been a dividing line between workstations/desktops and laptops/netbooks and their usage. As long as this line remains, there's little reason for all SSD to become nano-drives. And 1 and 2TB SSDs have been around for quite a while. Just not within reach of the home crowd.

 

[profdlp]

Excellent points, carwiz!

 

[Night Hawk]

"it's not going to be viable to go past 6.5nm ... 2024 is the end."

hmm, there are a whole 12 years of technological development to go until then.

in three years time, someone will invent an ssd super-duperlizer, then in five years that will get superseded by a mega-wegalizer etc etc - probably.

would you have listened to hard drive manufacturers a dozen years ago if they had said "well, 5 gigs looks like the limit - we'll never get past that.", or ram manufacturers saying "we're never going to be able to make a 64 meg ram stick."

Exactly.

At my first computer job I remember my boss (who was a really sharp guy) saying that 4GB was the absolute top due to the limits of FAT drive structure.

Then some bright lads created a drive structure with two FAT tables.

MS originally went from 8bit to 16bit and then onto 32bit for the Dos factor. One old case while everyone was pointing out the 74gb limitations I ran a single 234gb primary on a 250gb drive for 98SE. Before that I had even used a bios overlay program you had to see go on first before a regular format.

The newer form of exFat is what you now see on many external hard drives to compensate for the increased capacity over the years. For strictly Windows use however the recommendation when buying a drive is simply to reformat to NTFS from the start since that is a more secure file system.

I think SSD will be supplanted by some new faster technology before HDD ever catch up in any way. So in that sense they are doomed.

I wouldn't be surprized in the least while lately someone had posted information on a PCIe type SSD drive showing how manufacturers are looking for any other available space that can be used for drives.

The high price tag for the newer technology is nothing new either while seeing prices remain high is a deterrent to seeing them adapted by the common user on a budget!

As far as ATA standards eventually there will be a peak where you wouldn't be able to go any further. That's been a known fact for years now and likely one reason why SSDs were developed as an alternative.

As far as increased capacities over time I know some who would simply run several small drives rather then going with a few large capacity drives. But 25tb is still off in the distance regardless of the drive technology used at this time.

For any phenomenal change in seeing super capacities you would have to expect a new technology to be introduced. That would need to be a performance orientated as well otherwise face drive access time problems for huge amounts of data. That's the main drawn with having too large a capacity and something manufacturers would have to overcome.

 

[profdlp]

...The high price tag for the newer technology is nothing new either...

That's for sure. I always felt like I was the last guy in town for everything - VCRs, Audio CD Players, DVD Players, Flat-Panel Monitors, Flat-Panel TVs, you name it. They were all like a thousand bucks or more when they were "cutting edge".

For that matter, I remember the day I took my tax refund into Computer City and bought four 16MB sticks of EDO RAM. They were on sale for $99 - EACH! :shock:

(But I was the envy of the neighborhood, for about a month at least.)

 

[Night Hawk]

At least you had one thing you could say back then! Likewise I was the first running XP then Vista and now 7 while some people I know were late for XP upgrading from 98 and remain with XP except running Vista. Then they are stumped when learning a new board at this time may not see any XP support.

Here I've been waiting to see drive capacities increase as well as lower prices for SSDs to give one a try. Unfortunately we know the usual for that as far as any hard drive technology when you look back at the $200 retail tag seen less then 10yrs. ago for a 250gb ide drive no less and compare that pricing to a new SSD with 4 times the capacity!

A 500gb now catches about $45- being twice the size for 25% the cost of what was seen back then as prices overall have fallen. The SSDs however being the newer tech have to cover development costs as well as the large profit margin until prices are eventually forced down by the market itself.

The only other drawback with an SSD however over ATA mechanical is HD repair! Like a bad dimm what do you do with an SDD that develops a problem? Likely end up tossing it like a bad stick of memory!

With an ide or sata drive developing bad sectors a few programs like HD Regenerator can often save a worn drive provided the wear and tear isn't too severe. For many however with prices having come down a new drive is the usual choice for that however.

 

[Dave76]

The rumors of SSD lack of durability has been exaggerated.
SSD Write Endurance 25nm Vs 34nm
When a SSD fails due to wear it can't be written to but can still be read.
As with any device there will be failure exceptions but, this is generally accepted and the intended design strategy.

The average desktop user writes between 7-10 GB worth of information per day.
You can figure out your average writes.
On the above link, there were several SSDs around 500 TiB writes (550 TB) and still going. You can do your own calculations.
The easy one:
3.65 TB per year
550TB/3.65 = 150 years

Writes of 100 GB per day still gives you 15 years, not bad.

This is a graph I made in October:

 

[symp]

I hope,
They just start making Raid ssd drives, and no just those expensive onboard pcie cards.
So you end up buying a SSD drive not to much bigger then it is now, not bigger then a standard HDD that is like 4x128gb onboard in a Raid 0.
For people like me,
for the other people they could do block drives or something, takes up 2 slots has 10x128gb raid 5 built in, etc.

 

[Night Hawk]

A RAID array is two or more identical drives working together whether ide, sata, or now SSDs. Mirrored arrays, Striped arrays, Spanning volumes across multiple drives all require identical drives working in sync with each other.
The high prices along with the most often seen lower capacity models would be expected to see fewer arrays to begin with at this time.

As for durability over time Dave76 had an interesting point when you look at old systems that are running that used SIMMs not DIMMs for example PC100 or PC133 memory on an old Legacy Windows build shows that form of memory has lasted all those years. Will SSDs last longer then some articles would downplay?

The same can be said for any other type of drive which depends on the wear and tear and environment in which a drive is subject to. High temps for example will have an effect over time on any hardware which includes hard drives of different types. Besides temps you can look at handling of either drive or the system a drive is installed into.

You certainly wouldn't want to see one dropped on the floor! And as far as data still being read from a failed drive that would also apply to mechanical as well as SSD simply why so many data recovery services stay in business in that sense from being able to retrieve data from the platters inside of a PATA or SATA drive.

The one advantage of the SSD to add here however would be the lack of need for opening up the casing in order to access data from one that suddenly went out of commission. There are no drive platters to hook up under special conditions in order to rescue data.

Now for the actual disadvantages of data retrieval from SSDs they actually have more vulnerabilities then people are even aware of. One page seen at a data recovery service outlines just what those are. Solid State Drive Recovery | SSD Data Recovery

There are some other blogs on SSDs you may have or have not seen. Data Recovery for flash SSDs articles StorageSearch.com

The blog writer emphasized caution showing there will continue to be mixed opinions on any recent if not brand new techology as always. At the same time even pro recovery services are not always able to rescue everything from a failed mechanical drive. You still have to leave room for a margin of error.

 

[lehnerus2000]

I'm not sure about that

The one advantage of the SSD to add here however would be the lack of need for opening up the casing in order to access data from one that suddenly went out of commission. There are no drive platters to hook up under special conditions in order to rescue data.

Sure SSDs have no platters.

How do you access the memory chips if the controller has c****** itself?
Surely it will have to be opened up to allow access to (or removal of) the old controller.

Now for the actual disadvantages of data retrieval from SSDs they actually have more vulnerabilities then people are even aware of. One page seen at a data recovery service outlines just what those are. Solid State Drive Recovery | SSD Data Recovery

I do agree with this though.

I read an article about "data deduplication" on SSDs, which indicated that your data is actually more vulnerable!
Obviously that doesn't just apply to SSDs though.

 

[symp]

Raid Array being more then 1 drive,
doesnt have to be..
they could put a controller inside a drive, and run multiple layers of the IMCs (?) .
Arent they already doing that on Pcie based SSD's?
That would produce much faster drivers and more space on them without taking up pcie lanes.

Sure you probably wouldn't be able to raid multiple types of these drives together, but you wouldnt need to.

 

[jimbo45]

Hi everyone
As laptops become HUGELY more powerful now -- the old Desktop is really going the way of the DODO -- apart for a few niche markets like a few people who build their own gear, or need extreme gear for specific purposes like serious gaming.

A cheapish laptop with 8 or 12 GB of memory in it with at least an i5 processor will do pretty well ANYTHING the average computer user either needs or requires and will do it quickly too.

I don't forsee the need for RAID at all at the non server end of the market. Fast massive HDD's YES -- but if I'm simply playing a movie either from local storage or the Internet bog standard HDD's are fine -- and the newer USB3 devices are absolutely fast enough.

For loading and running the OS an SSD makes sense --also for running applications like photoshop which use a lot of "scratch data" for saving Layers etc --- If you ever process at least an A3 size -- or A2 poster size professional photo image you'll get the point pretty quickly.

However Hybrid disk are appearing which could make sense and eventually the OS will boot from programmable ROM again in my view making the long term future for SSD's look bleak.

The only possibilities that could change this outlook would be POWER consumption and PRICE -- if these devices can use minimal power and cost roughly the same as a conventional HDD then "All bets are off" as SSD would then replace HDD.

It's the scientists and Engineers who will ultimately settle this issue.

Cheers
jimbo

 

[seekermeister]

jimbo45,

It's the scientists and Engineers who will ultimately settle this issue.

No, it will be the businessmen paying those scientists and engineers that will settle the issue. It all about profitability.That means that indirectly, it is the consumers that makes the decisions.

 

[jimbo45]

jimbo45,

It's the scientists and Engineers who will ultimately settle this issue.

No, it will be the businessmen paying those scientists and engineers that will settle the issue. It all about profitability.That means that indirectly, it is the consumers that makes the decisions.

Hi there
actually this isn't as simple as it sounds -- Engineers these days HAVE to realize costs are a very important part of the process and the art of this type of successful engineering is to be able to forsee that the future costs will be significantly less for a SUPERIOR product which will have significant consumer or industry appeal..

All businesses require capital and whether it's Airbus 381's, Appollo Rockets or the individual selling burgers outside the Colliseum in LA they need to approach someone and "sell" a business model of some type in order to raise the cash.

The business model being sold might be accurate or total B/S but the sales pitch HAS to be done in such a manner that the bozos ponying up the cash believe it enough to start investing in it.

It's been a long long time since an Engineer just concentrated on "The Engineering".

Actually injecting some cold reality into this stuff keeps research basically on track and IMO isn't actually harmful to development either.

What IS really harmful is the incredibly short term attitude some CEO's make regarding their business -- chop everything so the share price looks wonderful for about 2 days and they walk away with a huge fat bonus. 6 Months later when there isn't anything left in stock to sell or staff left to make anything either and the company goes down the tubes that same CEO is performing his michief somewhere else on the back of a resume which says how he "saved" company XXX zillions of dollars and how the stock price rocketed --forgetting to mention it was only for 3 days.

I'm not anti market -- you DO have to adapt to changes etc but some of these CEO's should not EVER be allowed anywhere near ANY corporation EVER AGAIN.

Cheers
jimbo