How long before SSD overtakes Mechanical?

[Cogman]

So I just read the anandtech review of the SSD drive and I am really quite impressed. It looks as if SSD drives have finally caught up with regular Hard-Drives as far as raw transfer speed goes. Given the rate of memory development, I can't see the day being to far away when they overtake even the fastest of Hard-drive solutions and become standard for HD storage.

So where do you see them in the future, right now they are in a high end price range. (But remember, they have dropped from $100 per GB, to $33 per GB in just a year). My prediction goes something like this

Now, High end enthusiast -> Next year, Upper Mid level Enthusiasts -> Year after, Mid level Enthusiasts -> Year after that, Most new builds will have one.

So after 3 years I could easily see them overtaking current Hard drives. How about you?

 

[woolfe9999]

I too am pretty excited about SSD's. It seems that when this technology reaches maturity, there is no metric by which they will not be superior to mechanical drives, be it price, noise, heat, power consumption, or lifespan.

I think your timeline is about right. I would relegate its current status to ultra-ultra high-end, and only those who buy one just to have one. The problem is not just the $1000 price tag for the Samsung drive. It's that the absolute amount of storage space is not adequate for most gamers, and gamers make up the bulk of what we call enthusiasts. You have to store your games on the OS drive, and 64gb just doesn't cut it. I could exceed that with 6-7 game installs + the OS. It's no help that I could use a larger capacity mechanical drive for raw storage.

I'd say next year with the 4th generation drives, you'll see 128gb models at the $1000 price point, and high-end enthusiasts will be more likely to jump in at that time. The following two years will be trickling down to lower levels of enthusiasts. That puts 2012ish as about the time when OEM systems aimed at average consumers will routinely incorporate them, while they may be standard on laptops about 1-2 year prior. It's hard to say though, as the technology could advance slower or faster than whatever seems reasonable at this moment in time.

- woolfe

 

[semisonic9]

I don't think it'll happen until SSD can replace current drives in both price, availability, and performance.

Like the review said, I think they'll be relegated to specific notebook applications (tough-books, air-books, etc) and even more specific government/healthcare related uses for the immediate future.

Picking numbers out of a hat...2012 sounds good. Of course, if you're into the whole Mayan thing, that's basically the same as saying "the end of the world", according to some people. :-D

~Semi

 

[aka1nas]

18-24 months for enthusiasts on the desktop(ousting the niche the Raptor currently occupies). They'll still be using standard hard disks for bulk storage, though.

3-5 years for it to replace hard disks in the mainstream and value segments.

 

[ShadowFlareX]

I agree with aka1nas, it's still a long way off until even the enthusiats start going for these SSDs. It's just still way too expensive.

I'd love to have these someday in my system, no more HDD noises, much less heat, less power consumption, the lot!

 

[tcsenter]

Originally posted by: woolfe9999
I too am pretty excited about SSD's. It seems that when this technology reaches maturity, there is no metric by which they will not be superior to mechanical drives, be it price, noise, heat, power consumption, or lifespan.

It will be a good while before SSD can match mechanical drives in this respect: Data recovered from Columbia hard drive

Fires pose the most significant problem for mechanical hard drives, but drop from 10 stories, throw out the window doing 75MPH on the interstate, submerge in water for several months, collapse a building on it, or zap with errant currents - the odds of recovering that data are very good.

In addition, the logic-driven nature of the integrated circuit as a storage device poses significant technical challenges or limits to data persistence and retrieval should a critical portion of the circuit be damaged, requiring extraordinary means at extraordinary cost, if technically feasible at all.

Mechanical drives will be the only option for many years where data recoverability in the event or failure or damage is a priority.

 

[Brovane]

I don't see SSD catching up to mechanical hard drives any time soon.

I see a great potential for this technology in the Enterprise reals for SAN's and servers. You can have your OS drives for servers using SSD with a RAID 1. Since the SSD has fewer moving parts I would see the potential for greater reliability for serves. For SAN's you could have some DAE were storage space is critical setup with mechanical drives. Other DAE's on the same SAN can have SSD drives. Of course it will remain to be seen when SSD's can match the performance of 15K RPM fiber channel Disks.

 

[Cogman]

Originally posted by: tcsenter

Originally posted by: woolfe9999
I too am pretty excited about SSD's. It seems that when this technology reaches maturity, there is no metric by which they will not be superior to mechanical drives, be it price, noise, heat, power consumption, or lifespan.

It will be a good while before SSD can match mechanical drives in this respect: Data recovered from Columbia hard drive

Fires pose the most significant problem for mechanical hard drives, but drop from 10 stories, throw out the window doing 75MPH on the interstate, submerge in water for several months, collapse a building on it, or zap with errant currents - the odds of recovering that data are very good.

In addition, the logic-driven nature of the integrated circuit as a storage device poses significant technical challenges or limits to data persistence and retrieval should a critical portion of the circuit be damaged, requiring extraordinary means at extraordinary cost, if technically feasible at all.

Mechanical drives will be the only option for many years where data recoverability in the event or failure or damage is a priority.

Actually, Solid state drives are VERY resilient. Not only can you do just about everything to them as you can to a mechanical hard drive (Ok, high electric current would be the breaker) but you can plug them in right after you do it and get the data back.

Remember the Tsunami where they recovered a SD-Card from a camera phone that still worked, while showing video of the Tsunami hitting? True, a Hard-drive might survive that, but would you be able to plug it in again after it happens (or do you need professional services for the recovery).

But yeah, most of those things you posted are no problem for SSD drives either, in fact, I would say they have mechanical drives beat in that area. (heck, go look at the video of the guy boiling his Corsair Memstick).

The other thing that is nice about them is if you really need the drive really protected, you can put whatever type of casing you want around it (the little black box?) and it will still work without flinching, because it generates MUCH less heat then a hard-drive.

But yeah, I don't plan on dropping my $1000 SSD hard drive off a building any time soon 🙂

 

[tcsenter]

Originally posted by: Cogman
Actually, Solid state drives are VERY resilient. Not only can you do just about everything to them as you can to a mechanical hard drive (Ok, high electric current would be the breaker) but you can plug them in right after you do it and get the data back.

Ehh...you can always use them if they still work, just like hard drives. There are lots of dead SS devices after being subjected to lesser conditions, including those that up and die for no apparent reason (just as hard drives do).

Cell phones are designed to be fairly resistant to long term humidity, corrosive sweat and skin oils, which also gives them reasonable protection during short periods of water submersion (e.g. a few days max). The same could generally be said for portable devices that use mobile hard drives such as notebooks and video cameras. Although your notebook or video camera might not work after dropping it in the lake, or may require extraordinary measures to dry-out, you should be able to remove the hard drive and plug it after ordinary drying measures.

More rugged enclosures and better protective circuitry does nothing more for SSD than it does for hard drives. Ruggedizing is a preventative question. I'm talking about actual failure or damage. It is not enough to say SSD is a lot more reliable because it has no moving parts, unless "a lot more" means "never fails and can't be damaged". As long as SSD can and does fail, and can be damaged, then statistical reliability isn't an answer to recovering the data in the event of failure or damage and how much it will cost.

If the interface, electronics, or any other internal components of a mechanical hard drive fail or are damaged, you have direct and relatively uncomplicated access to the ones and zeros as long as the platters were not destroyed. If your SDD no worky after removing it from its protective casing and rigging a new interface, its game over.

Well, not definitively in all cases. There will be some cases where the damage can be repaired or bypassed, but it will require extraordinary measures with a substantial shift in the economics of data recovery relative to mechanical drives (e.g. say...$500 per GB instead of $50 per GB).

 

[Griswold]

One thing that is almost never mentioned is the potential abysmal write performance of current SSD due to the nature of the process of flashing memory cells. If the cell is empty, you'll see those cool 0.1-0.5ms times and amazing throughput. If the flash cell is containing data and/or data has to be manipulated, the content of the cell has to be read into a buffer in chunks of 16kb, altered and flashed back to the cell - in worst cases, this can take up to 200ms. That wont happen to any HDD currently available.

While that may not be a concern for system disks, and probably can be solved by smarter write policies, it is what makes SSD fall behind conventional HDDs in many scenarios outside the "system disk" world, right now.

 

[taltamir]

considering that intel claims it has a new SSD chip (controller or storage?) that it developed in house and does 200 MB/s and will be released within months... well it would completely annihilate regular drives in terms of speed.
Price is RAPIDLY dropping, so it should be in the 'high end" by next year...

And SSD are inherantly significantly more rugged then a regular hard drive. And as for recovery from atmospheric reentry damage...
1. That is why you upload ALL the date before reentering.
2. They can still use magnetic spindle drives...

I doubt they would though, SSD are just more rugged and durable.
Seagate rates its regular hard drives as able to survive a 300G impact.
OCZ rates it's SSD as able to survive a 1500G impact.

The SSD would thus fare much better at withstanding a drop from high altitude. (being able to survive an impact of 5 times the force).

And the storage and operating temperatures of SSDs are amazing!

 

[dv8silencer]

What about the problem of how many times an SSD can be written to? I thought they have limitations on the # of times it can be written to before they fail?

 

[imported_Scoop]

I won't buy a 32GB >500 euro SSD when it's not even outperforming the <100 euro 640GB WD in every situation. But looking forward to when this happens, then I'm interested.

 

[pugh]

Originally posted by: Scoop
I won't buy a 32GB >500 euro SSD when it's not even outperforming the <100 euro 640GB WD in every situation. But looking forward to when this happens, then I'm interested.

agreed

 

[taltamir]

Originally posted by: dv8silencer
What about the problem of how many times an SSD can be written to? I thought they have limitations on the # of times it can be written to before they fail?

It is high enough (and controllers do write leveling) to last many many years.
MTBF of SSD 2 million hours.
MTBF of high end drives like raptor and seagate 1.4 millon
MTBF of low end drives like all other WD drives 0.3 million.

Originally posted by: pugh

Originally posted by: Scoop
I won't buy a 32GB >500 euro SSD when it's not even outperforming the <100 euro 640GB WD in every situation. But looking forward to when this happens, then I'm interested.

agreed

Smart notion, but it is 500$ not 500 eur.
It currently outperforms it in MOST aspects, but not all. But the next gen coming in the next few months should outperform it by a LOT in all aspects, and go down in price some more.
Either way, for all but a few people waiting is in order. I expect to buy a 200$ SSD that more then doubles the speed of any mechanical drive in a year.
And for mainstream adoption to follow the next year.

 

[dv8silencer]

Originally posted by: taltamir

Originally posted by: dv8silencer
What about the problem of how many times an SSD can be written to? I thought they have limitations on the # of times it can be written to before they fail?

It is high enough (and controllers do write leveling) to last many many years.
MTBF of SSD 2 million hours.
MTBF of high end drives like raptor and seagate 1.4 millon
MTBF of low end drives like all other WD drives 0.3 million.

Well the reason I was asking is I didn't think the MTBF included the "# of writes" problem into its calculation, otherwise I wouldn't have asked 🙂
But thanks for the reply

 

[tcsenter]

Originally posted by: taltamir
And SSD are inherantly significantly more rugged then a regular hard drive. And as for recovery from atmospheric reentry damage...
1. That is why you upload ALL the date before reentering.

When traveling through the ionosphere, both coming and going, there are periods where comm and data transmission can become highly unreliable. Part of the data on Columbia's hard drive was recorded during this transition between gravity and microgravity.

I doubt they would though, SSD are just more rugged and durable. Seagate rates its regular hard drives as able to survive a 300G impact. OCZ rates it's SSD as able to survive a 1500G impact.

There is a different significance to "survivability" for SSD than for HDD. Both are measures/ratings of how much the drive can withstand and still operate normally. But hard drives have a second equally if not more significant survivability measure; payload survivability. IOW, how much shock/impact it takes to render the data unrecoverable.

An inoperable hard drive that is not terminally destroyed remains useful as a recoverable copy or backup of the data captured to it. The amount of impact/shock required to significantly damage the platters is many times greater than is required to render the hard drive inoperable. For SSD, these two survivability measures are far more likely than not to manifest in the same event; when the device no longer functions, the data will likely be rendered inaccessible or destroyed.

THAT is what I've been trying to point-out. You SSD fans seem to believe the only value in storage technology is the I/O benchmark or statistical reliability, while the data residing on the storage device itself is of no consequence. I didn't offer scenarios such as tossing a hard drive out of a moving car or collapsing a building on it as examples of things that mechanical hard drives would survive and still operate. On the contrary, I offered scenarios that would near-certainly render any hard drive inoperable but not necessarily useless or worthless.

Indeed, a dead hard drive can be far more useful or valuable than it ever was functioning, but a dead SSD will always be completely useless and worthless.

Well, maybe if it melted in a fire and took on some shape that resembled Elvis or the Virgin Mary. Then you could probably put it on Ebay and get many times more than it was ever worth.

 

[taltamir]

i see what you are saying, i just don't see any value in it... yea, if the hard drive breaks into peices, the platters would survive, and they can take a year to reassemble a drive out of it and get the data back like they did with colombia, if the SSD broke into peices, the data will be gone for good, since it is stored as an electric charge which will dissipate, rather then as a magnetized piece of metal.
And that is really assuming that the SSD broke, it takes much more to break said SSD then it does to break a drive. You don't know for sure that the SSD would be destroyed and the HDD survive... There are movies where people literally boil, freeze, and drive back and forth with a pickup on an SSD, you plug it in and it works, all the data is there AND accessible... would the data on a HDD platters be "potentially recoverable" after such a treatment?

Even if the platter has a better survivability then an SSD chip... SO WHAT?

The level of usefulness of such a capability goes beyond negligible. And you can just store do a RAID1 on an SSD and a platter drive if you really must...
Also... a 400MB drive in this day and age? it was probably an ultra rugged one... otherwise they have some serious issues.

All it really means is that NASA will keep on using platter drives for the platter surviveability, while the rest of us will use SSDs which are superior in each and every way, except recovereability of the data from a drive dropped on a commercial trip to the moon.

And this is all based on your ASSUMPTION that platters have greater survivability.
I am thinking of some scenarios... and as far as I can tell...
1. Run over by a car: SSD wins
2. Boiled in water: SSD wins
3. Baken in oven: SSD wins
4. Dropped from 300th floor / airplane / orbit, SSD wins.
5. Dropped in the ocean, fished out after a month: SSD wins...
6. Shot with a bullet: Both completely lost.
7. Melted to slug: Both completely lost.
8. Degaussed: SSD wins.

Now you say that if dropped from 300th floor / airplane / orbit the SSD will be destroyed and so is the data, and that a platter drive will break, but the platters would be recoverable. I think this is more up to chance, but that SSD's have a higher chance of surviving. But really, I admit to not KNOW for sure, it needs to be researched experimentally, but honestly, neither will affect my choice of drive purchase... EVER. (unless I am personally looking for a HDD to place in a space shuttle of my design)

 

[tcsenter]

HA! I love it when I'm wrong. Well...I don't love being wrong so much as learning something I didn't know.

After doing some digging, I must retract my statements that SSD 'functional' and 'data' survivability are likely to converge in the same event, that a dead SSD will always be useless and worthless, and that overall probability of recovering data from a dead or damaged SSD will be very low.

It seems that data recovery specialists like Kroll Ontrack have successfully recovered data from SSD that suffered some fairly interesting if not seemingly hopeless damage. They have actually repaired SSD flash chips that were literally blasted off their own wirebonds and solder pins, by meticulously laying down new wirebonds and solder joints. i.e. microsurgery for integrated circuits

However, I stand by my statements that, where recovery is possible for SSD, the economics of it will substantially favor mechanical drives due to the unique technical challenges and extraordinary measures required to repair damaged integrated circuits and manipulate logic-driven device states.

On the question of SSD market adoption, developers of mechanical drives have not depleted their bag of tricks. According to industry sources, a major unnamed hard drive manufacturer recently secured an order for specialized volume lithography equipment, suggesting it is almost ready to move patterned media recording layer technology out of the R&D phase and into the manufacturing phase.

At least two companies have successfully demonstrated 1Tbit/sq. in. areal recording densities using patterned media recording techniques. Industry papers indicate the technology should be mature enough to all but guarantee acceptable yields @ 500Gbit/sq. in. when first-generation drives enter mass production in the 2009/2010 time-frame.

To put that into perspective, Hitachi's newest 1TB 3.5" drives have a maximum areal density of 160Gbit/sq. inch using second-generation perpendicular magnetic recording techniques. Its 500GB 2.5" Travelstar drive has a maximum areal density of 260Gbit/sq. in. using fourth-generation PMR.

So it seems that hard drive companies will attempt to slow SSD trends by offering HUGE drive capacities with cost/GB that SSD won't be able to compete with for many years.

 

[Cogman]

Originally posted by: tcsenter
HA! I love it when I'm wrong. Well...I don't love being wrong so much as learning something I didn't know.

After doing some digging, I must retract my statements that SSD 'functional' and 'data' survivability are likely to converge in the same event, that a dead SSD will always be useless and worthless, and that overall probability of recovering data from a dead or damaged SSD will be very low.

It seems that data recovery specialists like Kroll Ontrack have successfully recovered data from SSD that suffered some fairly interesting if not seemingly hopeless damage. They have actually repaired SSD flash chips that were literally blasted off their own wirebonds and solder pins, by meticulously laying down new wirebonds and solder joints. i.e. microsurgery for integrated circuits

However, I stand by my statements that, where recovery is possible for SSD, the economics of it will substantially favor mechanical drives due to the unique technical challenges and extraordinary measures required to repair damaged integrated circuits and manipulate logic-driven device states.

On the question of SSD market adoption, developers of mechanical drives have not depleted their bag of tricks. According to industry sources, a major unnamed hard drive manufacturer recently secured an order for specialized volume lithography equipment, suggesting it is almost ready to move patterned media recording layer technology out of the R&D phase and into the manufacturing phase.

At least two companies have successfully demonstrated 1Tbit/sq. in. areal recording densities using patterned media recording techniques. Industry papers indicate the technology should be mature enough to all but guarantee acceptable yields @ 500Gbit/sq. in. when first-generation drives enter mass production in the 2009/2010 time-frame.

To put that into perspective, Hitachi's newest 1TB 3.5" drives have a maximum areal density of 160Gbit/sq. inch using second-generation perpendicular magnetic recording techniques. Its 500GB 2.5" Travelstar drive has a maximum areal density of 260Gbit/sq. in. using fourth-generation PMR.

So it seems that hard drive companies will attempt to slow SSD trends by offering HUGE drive capacities with cost/GB that SSD won't be able to compete with for many years.

Yay Competition! This is exciting news, basically a win for the consumer either way 🙂.

It seems as if the HD market has kind of stagnated at the 1TB max mark for the past year or so, This should be good news for Transfer speeds as well. Either way, storage should get much more interesting then it was in the past with just mechanical. (I have been somewhat unimpressed with the advancements made)

One things for sure though, SSD should take over mechanical some day. There are a lot of interesting SSD Techs out there that could really take it to the next level (Imagine your Hard drive being as fast as you ram, faster even, Heck, just get rid of the ram and use you're Hard drive 🙂) FRAM, PRAM, and MRAM all look promising.

 

[taltamir]

seems like next year i will own an SSD OS drive and a magnetic platter RAID5 array for my storage.. Multi terabyte drives sound delicious!

 

[LOUISSSSS]

just a thought, what if you can overclock these SSD drives? (its a stretch)

 

[Cogman]

Originally posted by: LOUISSSSS
just a thought, what if you can overclock these SSD drives? (its a stretch)

Bad Idea. like ram, an overclocked SSD drive could easily result in lost or corrupted data. I don't know if the speed boost would be worth it, maybe (Yay, HL2 loads 2 ms faster!)

 

[LOUISSSSS]

Originally posted by: Cogman

Originally posted by: LOUISSSSS
just a thought, what if you can overclock these SSD drives? (its a stretch)

Bad Idea. like ram, an overclocked SSD drive could easily result in lost or corrupted data. I don't know if the speed boost would be worth it, maybe (Yay, HL2 loads 2 ms faster!)

yea it was just a thought

 

[taltamir]

i think i saw once people overclocking both SSD AND mechanical drives (you can OC those too)...

but it is a really REALLY bad idea to do either