2008, The Year of Solid State Storage

An anonymous reader writes "At CES, SSD drives were a plenty on the show floor. "Some companies said we could see 250GB SSD units by the end of this year, while others predicted it could take up to a couple of years for them to become mainstream. None of the companies promised mainstream adoption, but they promised a bright future and we are inclined to believe them. High capacity drives are going to be expensive due to their very nature of early technology and gradual adoption rate."

Comments

Lets try the other way around, eh

[v1]

High capacity drives are going to be expensive due to their very nature of early technology and gradual adoption rate.

I think they have that backwards. Lets try High capacity drives are going to have a gradual adpotion rate due to their very nature of being expensive due to their being early technology

There, that's better.

I'd have one now ("be an early adpopter") if they weren't so bloddy expensive.

Re:Lets try the other way around, eh

[4solarisinfo]

Price isn't the only factor here. Has anyone seen any real reliability or Environmental numbers on any of these drives yet? I know many government/military programs who would be glad to pay for it, if it could prove to increase availability in certain environments.

Re:Lets try the other way around, eh

[minginqunt]

I remember seeing a curve of cost/gig over time of SSDs vs magnetic media, and it seemed to show that although both were falling, SSDs were falling faster, and were due to overtake their clicky brethren in the 2012-2014 time frame.

Once that happens, I imagine that magnetic drives' usage will tail off sharply, and disappear within a couple of years, because nobody (or at least nobody worth speaking of) wants to use magnetic over solid state anyway. In fact, it might start happening even whilst SSDs have a small price premium.

God knows, I'd be happy to pay a 20% premium to never have to use magnetic hard drives again.

Re:Lets try the other way around, eh

[JulieHo]

Tigi SSD 8GB costs around 20,000 USD. I am going to guess, this will hopefully be cheaper.

Re:Lets try the other way around, eh

[Geoffrey.landis]

Price isn't the only factor here. Has anyone seen any real reliability or Environmental numbers on any of these drives yet? I know many government/military programs who would be glad to pay for it, if it could prove to increase availability in certain environments.

Well, flash storage certainly is better in the space environment. Conventional hard-disk technology requires a pressurized compartment (the heads stay separted from the disks with a thin film of air). And, of course, any technology with no moving parts is preferable-- mechanical parts have an annoying tendency to freeze up with vacuum thermal cycling.

Spirit and Opportunity are now four years into their 90-day mission on Mars, running on flash storage....

Re:Lets try the other way around, eh

[Lars Clausen]

Just looking at newegg.com, I find the current sweet spot of SSD to be 32GB, at $250. $7/GB, half down from what Wikipedia mentions for "late 2007". The price is not just falling, it's plummeting like a jumbojet with both wings shot off. I love it. Can't wait to get the last mechanical pieces out of my computers.

-Lars

Re:Lets try the other way around, eh

[calebt3]

Can't wait to get the last mechanical pieces out of my computers What? Do they now have a CD drive that can read any part of the disc with no moving parts? Cool!

Re:Lets try the other way around, eh

[bunratty]

The cost for a given capacity will depend on the capacity needed. The smaller the capacity, the more advantage flash will have over hard disks. For now, 250 GB of flash is much more expensive than a 250 GB hard disk. On the other hand, you can get 1 GB of flash for under $10. Are there any hard disks at all available for that price? Also given that flash is faster, smaller, and consumes less power than disks, flash will replace disks in devices that need smaller capacities first. That means the usage of drives will decrease gradually from now until the 2012-2014 time frame you mention.

Re:Lets try the other way around, eh

[Lars Clausen]

What? Do they now have a CD drive that can read any part of the disc with no moving parts? Cool! No, but I don't expect my laptop to have an internal CD drive, for as rarely as I need it I'm fine with an external one. Also saves space and power, not to mention the internal "gotta spin up the CD just for fun" symptom.

-Lars

Re:Lets try the other way around, eh

[alexhs]

nobody wants to use magnetic over solid state anyway Oh, but I've heard that magnetic data has a warmth and nuanced feeling that SSD harsh data doesn't have...
Already that magnetic drives weren't all that good to start with...

Re:Lets try the other way around, eh

[keithpreston]

Yeah, but if you look at SD cards the raw retail cost of flash is about $4/GB (newegg) So realistically the 32gb drive is only worth 128 dollars in flash, plus the cost of the IDE/SATA flash controller logic. The problem is that this logic is new and probably expensive. I give it about a year for volume and competition to ramp up on these controllers, and then they will drop to commodity pricing. Then I would guess the sweet spot would be $2/GB for flash, 64gb drive (2x$64 + $10 (controller logic)) = $138. Give it a year.

Re:Lets try the other way around, eh

[cbreaker]

No, no.

Hard drives are actually vented. There's no pressurized compartment. They run at the same atmosphere as the rest of the machine. The lift of the hard drive heads is the "Bernoulli effect" (http://en.wikipedia.org/wiki/Bernoulli's_equation) see also (http://news.bbc.co.uk/1/hi/programmes/click_online/5413198.stm).

Flash storage is certainly preferable in hard environments, no doubt. But as far as I've seen, I'm not a convert. Useful as hell for digital cameras, PDA's, MP3 players and USB drives. Not so much for primary storage. They're simply not fast enough, usually. I guess some people are making faster ones, but you still can't affordably beat the ol' hard drive when it comes to transfer speed. Seek times, flash wins, but when the transfer rates are as slow as they are on most flash, it doesn't matter.

Re:Lets try the other way around, eh

[iminplaya]

It might be anecdotal, but this may shed some light.

Re:Lets try the other way around, eh

[jenik]

i think the previous post meant they need to be pressurised in space, since normally there's no air there.

Re:Lets try the other way around, eh

[Duncan3]

And don't forget the horrible write performance.

By the end of this year, they won't suck, and another halving of price will approach reasonable.

Re:Lets try the other way around, eh

[mlush]

No, no.
Hard drives are actually vented. There's no pressurized compartment. They run at the same atmosphere as the rest of the machine. The lift of the hard drive heads is the "Bernoulli effect" (http://en.wikipedia.org/wiki/Bernoulli's_equation) see also (http://news.bbc.co.uk/1/hi/programmes/click_online/5413198.stm).

So when these drives are exposed to hard vacuum (as suggested by the OP) the Bernoulli effect fails and the heads start gouging into the platters.

Re:Lets try the other way around, eh

[wwwillem]

What? Do they now have a CD drive that can read any part of the disc with no moving parts? Cool!

Yes, that's called a USB stick!! A one gig one (CD size) is nearly a give-away, and a DVD matching 4 GB stick is by now pretty affordable as well. And hey, you can even boot from them.

Re:Lets try the other way around, eh

[calebt3]

Last I saw, I couldn't pick up a 100-pack of 1-Gig flash drives (I haven't been to the flea market forever...) for $25.

Re:Lets try the other way around, eh

[BitZtream]

Bulk trasfer speed does not have to be slow. Its not like your hard drive in your PC now reads from one head at a time. It uses multiple heads to achieve higher rates by reading/writing all the platters at one time. The same applies for your RAM. You don't read 32 bits from a single chip in a clock cycle, your stick of ram has several chips on it, they all get strobed at once to return a larger size. Then there is the whole dual channel thing, not only do you do it with each chip on the stick, you do it with 2 sticks on seperate channels so you can now read in twice as much as a single stick.

So with flash memory you don't put in one really big chip to get 250GB, you put in 250 1GB chips working in parallel. Instant 250x increase in throughput using a relatively minor increase in die real estate for the extra controlling circuits. The only reasons USB flash drives are slow now is because A) they are dirt cheap B) no one is using them in a way that REALLY needs to be fast C) its on USB anyway, not like we're talking about a high performance bus in the first place

Re:Lets try the other way around, eh

[LooseBrie]

Can't wait to get the last mechanical pieces out of my computers. I'm intrigued how you read CDs/DVDs. I'm also interested how you keep your machines cool.

Re:Lets try the other way around, eh

[Anonymous Coward]

You magnetophiles really take the cake ... I suppose next you'll be telling us that if we can't tell the difference on our monitors between data pulled from a simple SSD as compared to your overpriced magnetic platter storage with wooden control knobs and monster cable connecting everything together, it's because our vision isn't discriminating enough.

Re:Lets try the other way around, eh

[tlhIngan]

I remember seeing a curve of cost/gig over time of SSDs vs magnetic media, and it seemed to show that although both were falling, SSDs were falling faster, and were due to overtake their clicky brethren in the 2012-2014 time frame.

That must've been a while ago... SSDs can only drop in price as fast as Moore's Law (only faster if someone dumps them) - the bytes/area of silicon is fixed by Moore's Law (as is the cost/area of silicon - or why full-frame DSLRs are always going to be pricey - silicon wafers are more or less the same cost, until some breakthrough means we can use larger wafers. But it's taken many years to go from 12cm wafers to 30cm wafers (decades). Hard drive storage seems to double faster than Moore's Law, practically doubling yearly. Last year's pricey drive was a 500GB one, and now 500GB is economical, with 1TB disks being pricey.

Of course, right now, the technology is immature, so as it matures, it may be increasing in space temporarily faster than Moore's Law (as we figure out how to do controllers to do accesses over multiple devices, etc).

Re:Lets try the other way around, eh

[cbreaker]

Well, the USB bus is capable of 480Mbit. USB overhead is quite high for high speed transfers but still, you could achieve 40MB/sec over this interface. It's not that slow.

Current consumer flash-based hard-drive replacements are still slow as shit. Yea, you could do it all crazy with 250 1GB sticks to achieve good performance, but those are already available and they cost HUGE DOLLARS.

I realize that eventually, Flash will catch up and could very likely replace hard drives. I think it sounds wonderful. But it's just not here yet, even with the new disks introducing this year.

Re:Lets try the other way around, eh

[Geoffrey.landis]

No, no.
Hard drives are actually vented. There's no pressurized compartment. They run at the same atmosphere as the rest of the machine. My error; I apparently failed to be sufficiently explicit in what I wrote. When I wrote "in the space environment," what I actually meant to say was "in the space environment, which is a vacuum, a technical word which means that there is no atmosphere..."

Re:Lets try the other way around, eh

[Real1tyCzech]

Well, no...

But he said "out of", not "no longer plugged into via USB". :p

Re:Lets try the other way around, eh

[Anonymous Coward]

Oh, but I've heard that magnetic data has a warmth and nuanced feeling that SSD harsh data doesn't have...
Already that magnetic drives weren't all that good to start with...

A musician friend of mine has said that he prefers the warm, nuanced sound of DAT to CD, and that it's because of the magnetic tape format.

Re:Lets try the other way around, eh

[Fweeky]

Its not like your hard drive in your PC now reads from one head at a time. It uses multiple heads to achieve higher rates by reading/writing all the platters at one time No it doesn't. Heads can only "focus" on a single track at a time due to the tolerances involved; you'll notice that single platter drives have much the same STR as drives with 4 double-sided platters, and STR increases are only associated with RPM and platter density.

Re:Lets try the other way around, eh

[amorsen]

Bulk trasfer speed does not have to be slow. Its not like your hard drive in your PC now reads from one head at a time.

Err yes, in fact, that's exactly what it does.

It uses multiple heads to achieve higher rates by reading/writing all the platters at one time.

No. It was attempted, I believe by Seagate in the first Barracuda drives, but it was quickly abandoned. The only way it can work at modern capacities is if you added a drive motor and independent electronics per head. Doable, but it's cheaper to just buy two drives and do RAID-1.

Your points about flash are correct though.

Re:Lets try the other way around, eh

[afidel]

For laptops that's great but for a desktop I'd just buy 2x500GB drives(RAID-1) for $180 or so and have MUCH better write performance =) Also unless you're running something slower than a Via C3 you still have those pesky fans to worry about, I've minimized mine but I still have 4 of them (one intake, one outflow, one PSU and one CPU).

Re:Lets try the other way around, eh

[drinkypoo]

Companies have experimented with having independently-moving heads, but I don't think anyone is still doing it. I do believe that there's still someone making hard drives with two sets of heads, but I think that's about as close as you get.

Re:Lets try the other way around, eh

[drinkypoo]

You had two options for correcting my post (which I thought it read "in a small space"): 1) Correct me by being polite (like some of the other posts) and assume I just read it wrong 2) Be a dick You chose option 2.

I would argue that they chose option 3) all of the above.

Once they did so you had two options.

1) Accept it and move on
2) Whinge

You chose option 2.

Re:Lets try the other way around, eh

[billcopc]

Yep that's what gets me too - flash is too slow for me to care right now.

My question is: why can't they just parallelize or something ? If my ghetto 1024mb Kingston USB stick can do 10mb/sec sustained, stick a couple dozen of those chips in a 3.5" drive-shaped box and max out the SATA bandwidth. Striping works for clunky hard drives, why wouldn't it work for flash ?

Re:Lets try the other way around, eh

[Joe The Dragon]

where are the firewire flash drives

Re:Lets try the other way around, eh

[Gyga]

No CD/DVD, usb drives for everything and download music/movies? Open air case left outside in the freezing winter?

Am I the only one still chugging along with a 20GB harddrive that cost me a lot of money?

Re:Lets try the other way around, eh

[galanom]

20% premium?

Ahem, I just bought tape cartridges (LTO4) for $45 and they have a native capacity of 800GB.
Exactly WHEN am I going to be able to buy an SSD drive of 800GB for $45? In 2012-2014? I really doubt it :(

Re:Lets try the other way around, eh

[cbreaker]

Hey, that's not original at all! All you did was take what I said, and put in your own words. How about a little creativity?

Bah, I don't have time for this. I have too much Whinge to do.

Re:Lets try the other way around, eh

[fenrisulfur]

"Four years into their 90-day mission" Priceless

Re:Lets try the other way around, eh

[dfn_deux]

The 32GB SSD you can buy for 250USD will be unlikely to provide any performance or reliability increase over what that same 250gb would buy you in traditional magnetic media. I've done perf/reliability testing on all the current generation SSDs from mtron, stec, memoright, and ritek. And even the 800-1000USD SSD drives fall way short of their predicted write lifetime when put into any environment where i/o is primarily small random r/w operations vs large sequential stream r/w operations. Write leveling is generally limited to 512b or larger blocks which makes it ineffective for these types of loads, furthermore the cheaper devices tend to use MLC flash nand vs. the less dense SLC.

Re:Lets try the other way around, eh

[darthdavid]

I love how slashdot mods can look at a scathingly sarcastic comment and mod it "insightful".

Re:Lets try the other way around, eh

[Jeppe Salvesen]

It's amusing how people who complain about other people's grammar always make a mess making their point. "bloddy". What exactly is that? Do I want to know?

Re:Lets try the other way around, eh

[Cyberax]

Right now, I have a computer with 150Gb SSD drive - it's a rugged industrial design for high-vibration environments. It works very fast, about as fast as 7200RPM desktop drive on bulk reads/writes and _much_ faster on random access.

There's only one problem: this SSD drive costs about $5000.

So we have the technology, we only need to wait until prices come down to reasonable values.

Re:Lets try the other way around, eh

[mandolin]

Hard drives are actually vented. There's no pressurized compartment.

Not always correct. From the wikipedia article on 'hard disk':

"Specially manufactured sealed and pressurized disks are needed for reliable high-altitude operation, above about 10,000 feet (3,000 m)."

Re:Lets try the other way around, eh

[SixDimensionalArray]

See slide 3 of former Microsoft Research Jim Gray's (RIP) presentation on flash drives for the diagram referred to in the parent. This is a great summary of what one of the most informed individuals believed about the future of SSD. I'm waiting! :)

http://research.microsoft.com/~Gray/talks/Flash_Is_Good.ppt

SixD

Re:Lets try the other way around, eh

[dintech]

Can't wait to get the last mechanical pieces out of my computers.

Building T-1000s are we?

So we are back to RAM drives!

[LWATCDR]

Oh well everything that old is new again I guess.
I used a RAM drive on my Amgia way back when. Yes I know that they are how using flash but it does seem very familiar.
I wonder when we might see a hybrid flash-ram drive? A big bunch of ram for high speed and flash for permanent storage. Just use a super cap for a power backup and have it copy the ram to flash on power down. A little bit pricey but if you need the speed you need the speed.

Re:So we are back to RAM drives!

[gEvil (beta)]

First thing I did when I put 64MB in my old Mac Performa 6100 was set up a 40MB or so RAM drive. I then copied the System Folder over to it and set it as the startup volume. That's still about the fastest I've ever seen a machine boot (took about 3-4 seconds from the startup chime). Too bad it wasn't non-volatile memory...

Re:So we are back to RAM drives!

[LiquidCoooled]

I have never seen the point in setting up a fixed RAM drive.
The amiga natively supported a variable sized RAMdrive which was always 'full' but as long as physical memory was available still had space.

Anything else since has seemed archaic.

Re:So we are back to RAM drives!

[gEvil (beta)]

I have never seen the point in setting up a fixed RAM drive.

I think it had more to do with how the old MacOS handled memory management. That is, very poorly.

Re:So we are back to RAM drives!

[Sen.NullProcPntr]

I wonder when we might see a hybrid flash-ram drive? A big bunch of ram for high speed and flash for permanent storage. Normal magnetic hard drives already do this to speed up sequential access (read ahead) among other reasons. No reason to believe this feature won't be transfered to SSD media. Although flash is much faster than magnetic media already.

Re:So we are back to RAM drives!

[Jeremi]

I wonder when we might see a hybrid flash-ram drive? A big bunch of ram for high speed and flash for permanent storage.

But isn't this pretty much what we have now for every drive we use? The only difference is that the "high-speed RAM cache" is located in the unused portion of your computer's RAM, instead of being part of the drive itself. I'm not sure what the advantage of putting another cache inside the drive itself would be; why not spend the money adding more RAM to your computer instead... that way the cache can be used by any disk you attach, and isn't restricted to just one drive.

Re:So we are back to RAM drives!

[k8to]

There can be issues of allocation in a multi-user environment, where you want to have a set ceiling for the drive. In some cases this ceiling has to be low, so there's not a lot of benefit of dynamic allocation over just allocating the dang thing.

Certainly in a personal-only environment where I do not compartmentalize security, it seems obviously desirable.

Just saying, it isn't always what you want.

Re:So we are back to RAM drives!

[ThreeGigs]

My 'ideal' hybrid drive would have DRAM, Flash *and* magnetic platters.

2GB or so of DRAM as a buffer and as a /temp filesystem for things like internet cache, swap and other stuff that can (and maybe should) be cleared/erased on a reboot. Flash for the hardest hit bits, like FAT tables and disk indeces so the heads on the platters don't have to jump around as much, and magnetic media because it's cheap.

Downside is, you'd really need an OS that could 'tag' writes as volaitile/io priority/standard data, and some smarts built into the drive itself to take advantage of the tags.

Re:So we are back to RAM drives!

[Heddahenrik]

ramfs and tmpfs on Linux work so that they don't take any memory when not filled with something, but they also have a specified maximum size.

As a side-note I can tell that putting a database on tmpfs is a really bad idea because it gets very slow when it's swapped out (compared to being on a disk in the normal way) and it will get swapped out. ramfs worked like charm, but it makes it hard to shut down the server cleanly, of course.

Re:So we are back to RAM drives!

[robosmurf]

RAM for a RAM disk could potentially use slower (and cheaper) memory than the main memory.

Also, as it wouldn't have to be attached to the main memory controller, there is the possibility of adding more RAM than could be supported by the main system.

Neither of these reasons is amazingly compelling, but there might be a niche for this. Particularly if people are still stuck with 32bit Windows, and thus limited in main RAM size.

Re:So we are back to RAM drives!

[ch-chuck]

I always setup a ram disk for swap space - it's so much faster than regular drives.

Re:So we are back to RAM drives!

[Tmack]

I have never seen the point in setting up a fixed RAM drive.
The amiga natively supported a variable sized RAMdrive which was always 'full' but as long as physical memory was available still had space.

Back then, in the days when ram was still $50/Mb, and HDs were just breaking the 1Gb range, I was still stuck in 386sx land with 8Mb of DIPs and a 60Mb MFM drive. To make things bearable, I installed Slackware Linux (my first Linux system!), which gave me 10Mb free, and loaded 10x faster than windoze 3.11 or 95 (not to mention I could actually run stuff on it). To load Netscape in X took about 6minutes still, so to speed things along and give me a bit more drive space, I tgz'd it and created a ramdrive that it was uncompressed to on boot. Load time for netscape reduced significantly, and became actually useable... until I attempted to install a 387 to the board, which only had solderpoints, not a socket, and made the magic smoke gods angry at me who then released their fury in a bright blue flash and then freed the smoke in one puff... Drive still works, and the 387 (missing one pin, that created the fash) is now hanging from my rearview mirror.

Anything else since has seemed archaic.

Yeh, guess it was pretty archaic...

Tm

Re:So we are back to RAM drives!

[LWATCDR]

It has been that way for a long time it is called a cache.
but the cache isn't as big as the drive.
Flash is actually slower for writes and has limited write cycles.
What I was imagining was using a ram drive for reading and writing data and then backing that up to a slow flashdrive when you powered down the drive.
On power up You could pre cache the ram or just use it as a very large cache.

Re:So we are back to RAM drives!

[pragma_x]

Downside is, you'd really need an OS that could 'tag' writes as volaitile/io priority/standard data, and some smarts built into the drive itself to take advantage of the tags.

I'm not an OS-driver developer, but that's nothing a custom filesystem driver for bsd/linux couldn't handle in some shape or form. You could probably base it off of file use patterns (allow for a 'learning' period before it gets smart) and have it do all that automatically without need for a specialized API or changes to how the OS works. Sprinkle a little etc/conf magic on it and you're good to go. :)

Re:So we are back to RAM drives!

[amorsen]

2GB or so of DRAM as a buffer and as a /temp filesystem for things like internet cache, swap and other stuff that can (and maybe should) be cleared/erased on a reboot

Why do you want to place the DRAM behind a slow SATA bus? Even if you switch to a better bus, nothing beats having a dedicated channel to the memory.

Re:So we are back to RAM drives!

[drinkypoo]

IIRC, while RAM: (the growable ramdisk) was dynamic, RAD: (the recoverable ramdisk) was not. The Amiga (like many architectures) did not zero memory on a reset; and the recoverable ramdisk could be remounted after a reboot! This let you copy portions of the OS to the ramdisk and then run them from there, and then not need to reload them after a crash - which was frequent, due to the lack of memory protection. Also, using a dynamic ramdisk can lead to memory fragmentation, and in order to use graphics or sound resources they must be loaded into contiguous sections of CHIP memory, because the custom chips (most notably Agnus and his descendants, which bit-blit) can only access CHIP memory - the first 512k, 1MB or 2MB depending on the model.

Re:So we are back to RAM drives!

[drinkypoo]

FPM DRAM costs more than EDO DRAM costs more than SDRAM costs more than DDR SDRAM, because they make less of the old stuff and more of the new stuff. Go on, check some prices! Sure, the old stuff is still around, but the capacities are low and it's distributed across the country, not in the hands of retailers.

Re:So we are back to RAM drives!

[Sen.NullProcPntr]

It has been that way for a long time it is called a cache.
but the cache isn't as big as the drive.
Flash is actually slower for writes and has limited write cycles. True, but flash chips in parallel (the way SSD are made) make that less of an issue. Sort of how certain RAID configurations can speed up disk access times. Samsung quotes maximum write speeds of SSD higher than equivalent magnetic HDD. Even the MTBF numbers are much much better for SSD. Of course the write speed is the maximum-guaranteed-never-to-exceed number the slowest write may very well be slower than the slowest HDD write.

What I was imagining was using a ram drive for reading and writing data and then backing that up to a slow flashdrive when you powered down the drive. On power up You could pre cache the ram or just use it as a very large cache. I see, that would be a very fast drive (once all of flash has been read into cache) and also expensive - how much does 160G of DRAM cost today? Prices will go down but disk capacity will probably go up even faster.

Re:So we are back to RAM drives!

[adisakp]

A big bunch of ram for high speed and flash for permanent storage. Just use a super cap for a power backup and have it copy the ram to flash on power down.

They already have this device (although it uses a battery rather than a supercap) in the form of the HyperDrive 4 and the RamSan. Supercaps are pretty darn huge (and expensive) compared to batteries for the same amount of energy storage and since the drain is relatively stable during the backup operation, there's no reason to use one over a battery. Both of the devices are rediculously expensive though (between $250 to $2000 per GB).

Re:So we are back to RAM drives!

[LWATCDR]

I was thinking of database servers as one use. Unless you are storing a lot of video, graphics, or audio databases are pretty small. A 160 gigabyte database is a pretty big database. But yes it would be very specialized and very expensive.

Re:So we are back to RAM drives!

[owlstead]

I still use a RAM drive, right out of main memory. I use it to unzip files from, for logging during performance tests, various temporary files and even for larger build environments nowadays. Nothing beats a RAM drive to compile to/from.

It's currently only 64MB which is enough for many applications, but I am probably going to increase it to ~ 500 MB, big enough for most archives etc. I might even buy more main memory just to set it at the maximum size (2 GB or such, I'm still at 32 bit) so I can download to it without spinning any drive (and disturbing my development and/or gaming with continuous disk I/O).

Memory is relatively cheap nowadays. Use it.

Re:So we are back to RAM drives!

[darkpixel2k]

I remember back in "the day" I would occasionally setup a 2 MB RAM drive, throw a bunch of files on it, delete a few, put more files in, append to a few, delete a few more, and really really fragment the heck out of the RAM drive. Then I'd load the old Norton defrag tool and watch in amazement as it defragged everything in under 0.68 seconds.

Of course nowadays I just run Linux which doesn't have fragmentation issues...the magic is sorta lost.
*sigh*

Re:So we are back to RAM drives!

[k8to]

Yeah, it's true, ramfs and tmpfs both work this way. But they've not seen adoption because on a modern unix with a backing buffer cache where short-lived files don't necessarily ever get assigned backing store, the advantage of using a ram-based filesystem at all is pretty ephemeral.

within 5 years, tape manufactuers will have troubl

[WindBourne]

HDD still have a ways to go. In particular, the flash storage will be used for desktops, laptops and the core of servers. The real data will still reside on HDD for a long time to come because of cost / MB. What will happen is that HDD will learn to really park and lower their energy needs, most likely due to dropping in size. Tape has been used for eons for back-up, but I think that HDD will overtake that role as their prices will be forced to go way down.

apple

[theMerovingian]

The sales guy at the Apple store told me that there was a persistent rumor of a solid state laptop coming in the next few weeks...

Boot camp + solid state = me finally replacing the old powerbook!!

Re:apple

[PrescriptionWarning]

they'll tell you the sky is purple if you'll only buy a Mac because of it

P.S. My Mac Mini rules

Re:apple

[Ryukotsusei]

Apple? Rumor? Thinksecret has been shut down! There are no more rumors to be had!

Re:apple

[SuperStretchy]

It might even be the new mini-notebook people have been talking about.. It makes sense

Re:apple

[Tibor the Hun]

So you're assuming that windows would just magically compile the drivers for all the new hardware they may be introducing in their ultraportable?
For all we know that thing may well have a multitouch screen. What use would your bootcamp be for that then?

Unless the windows table version is included into whatever version of windows you're intending on running on the new ultrap.

Re:apple

[FuturePastNow]

I guarantee you the sales guys at Apple have no knowledge of future products.

Re:apple

[andrewd18]

They'll tell you the sky is purple if you'll only install Linux because of it.

Re:apple

[Anonymous Coward]

The sales guy at the Apple store told me that there was a persistent rumor of a solid state laptop coming in the next few weeks... The "sales guy" is behind the times. Solid state laptops are already here from Dell, Sony, Fujitsu, and others.

Oh... you meant an Apple laptop. Well, Apple is late.

Sequential reading?

[ookabooka]

I was talking to a gentleman from (big name hard drive company) about their plans for hybrid and/or solid state drives. Essentially he told me that solid state was still limited by price and sequential reading. So it may be advantageous to put some things on flash like OS files that require a lot of random seeks, but for sequential reading of things like media files, traditional hard drive tech won't die just yet . . .I apologize for being too lazy to back this stuff up with numbers, what can I say, I'm a true slahsdotter.

Re:Sequential reading?

[jdunn14]

I've never seen the performance numbers for sequential vs random read on flash drives, but you have to do pretty damn bad to get beat by random access on a standard hard drive. If you look at the the units used you'll get the idea. Your average random access on a standard drive is based on the average seek time which is measured in small milliseconds (4 ms, 8ms). Access time for flash drives is measured in double-digit nanoseconds (e.g. 60ns). That's 5 orders of magnitude difference. Even if the access time for random reads on flash was 100 times worse than it's average access time those reads would STILL be 1000 times faster than from a hard drive.

I don't think people realize just HOW slow drives are compared to the rest of the machine. Sure we programmers know the disk is "slow" but it really puts it in perspective to know it's a 100000 times slower than an alternative tech.

Re:Sequential reading?

[Anonymous Coward]

For sequential reading a SSD would still SMOKE a HD. As that is what they were designed for. Most experiance people have with SSD is thru a USB 1.1 interface. Which is not exactly that fast. Also random with the type of flash currently being put in these devices actually has a random performance hit. Also by putting small chunks of S/DRAM they can smooth out any speed bumps. You can quite litteraly put a gig of DRAM memory on the thing for cache.

Where traditional hard drive tech has the SSD is price per MB. You can build a 1TB SSD but it will cost a bunch currently. At least WAY more than the $~300 1TB HDs are right now. You are seeing about $10 per GIG on SSD drives currently. That is pretty high up there vs the 0.30 for HD.

Re:Sequential reading?

[kestasjk]

Well those are seek times, as you said. Reading/writing continuous data is very fast, and the OS (and some HDDs) will use memory caches so that data access will be continuous as possible. The problem of hard disk seek times has become less and less of a problem as memory has increased.

Re:Sequential reading?

[Anonymous Coward]

The OS certainly lower the number of seeks by combining writes and can improve the average seek time by reordering writes so the heads are only moving in one direction, but it can only do so much. For many database applications, there is SO much random access that seek time dominates everything. Being a database, you want the data written to disk quickly, so you don't lose any data, and you may even want it written to the physical media in the order it was queued to maintain data consistency. In my usage, the data set is small enough that a 32gb SSD would be more than enough, and would give orders of magnitude speed improvements due to the huge amounts of random writes.

Re:Sequential reading?

[Fweeky]

SSD's do rather poorly for random writes; even a decent MTron drive's only rated for something like 130/second, and for the cheaper ones aimed at portables you can drop a zero off that. To actually get decent write performance you have to start jumping through hoops to convert the random writes to sequential.

Re:Sequential reading?

[Chirs]

"The problem of hard disk seek times has become less and less of a problem as memory has increased."

That assumes that you can cache your accesses. There are applications (various databases, for instance) where the working set is too big and accesses are too random for caches (even quite large ones) to help much. For these sorts of applications, solid state drives are very interesting.

Re:within 5 years, tape manufactuers will have tro

[DeeQ]

I don't think so, tapes will still have many uses. They are very reliable for backups more than I would trust a hard drive. They also have the ability to be taken to off site locations like all backups should be. Hard drives would make doing that a little more difficult, even with external Hard drives it would be more of a pain than having the media of a tape. I don't see the tape going away any time soon.

I dont see it

[Rooked_One]

Maybe in 09... not 08... unless we get chipsets that can supply greater throughput, the chipset will become the bottleneck - therefore, the only reason to have one of these is in a laptop or desktop... and thats for people for whom price is no object.

In the enterprise sector... forget about it... Even SATA drives are becoming ideal for storage solutions, and a simple raid-5 will max out the cap of a raid controller's bus.

So in other words... I don't see it.

Re:I dont see it

[Lars Clausen]

We have already been running tests showing Lucene to be several times as fast on large indexes and realistic queries using SSD than using normal drives. I'm going to have a smallish SSD in my new laptop combined with an external drive for my large data. Faster, more solid, and less battery usage. Doesn't matter if I get 32GB rather than 160GB on board. I agree fully with the OP, SSD will really break through in 2008. Dell already offers it as an option. It's all a matter of usage patterns right now, in the long term I am prety sure hard disks will die.

-Lars

Re:I dont see it

[Rooked_One]

I fully agree - but not in 08. As far as large indexes and such - raid-10 yourself and save some duckets. I have first hand exposure to this stuff and for 99% of people, they are wow'ed by the performance of a simple SATA raid array.

Re:I dont see it

[Lars Clausen]

I would give you the data on the direct comparison, but they haven't been published yet. Even over RAID-5, we get twice the speed without any warm-up time. And once you're talking high-performance HDD + RAID controller + extra disks for RAID + extra power for HDD + extra power for cooling, the savings on HDD are minimal if not gone.

-Lars

Re:I dont see it

[initdeep]

I have a 64GB SSD Drive in my Dell M1330 Laptop. I too use an external as storage for files.
The differences, side-by-side, to one without it, simply for OS Startup, are easily 3 to 1 in speed.

I was lucky enough to basically get the drive for free due to the EPP program coupons and discounts and other discounts..
otherwise i would never have gotten it.

But I'm sure glad i did.

I've also noticed a slight increase in battery life, although this could be simply a small difference in batteries themselves.

Re:I dont see it

[NSIM]

Your ignoring a number of issues. 1. Many enterprises are not facing a capacity crunch, they are facing an IOPs crunch where they simply can't get enough I/Os out of disk without deploying large numbers of them with low capacity utilization to get the benefits of multiple spindles. 2. Energy consumption is a big problem in many data centers, the first target to reduce that has been the server farms, but storage is next on the list (probably becomes the number #1 energy consumer in the data center during 2008-2009) Disk is by far the biggest component of overall storage power consumption. 3. There is nothing that can be done with conventional HD technology to impact either (1) or (2) Ergo, enterprises will have to start looking at alternatives and Flash is the most mature around at the moment. Evidence for this can be seen in EMC's announcement today of flash drives for it's DMX storage arrays.

Re:I dont see it

[Rooked_One]

i agree - in a round about way. while dell and others are starting, the idea is just emerging... these drives will make for good next x-mas gifts, but right now, too expensive for your average joe.

I work with a major computing company's storage division and SSD is no where in the talks right now which is pretty much why I state what I am. ;)

and to be completely honest, in the last year, only one person has asked about SSD. Seriously. And I talked to probably around 2000 people who use sas and sata storage solutions.

Re:within 5 years, tape manufactuers will have tro

[Anonymous Coward]

That's hardly the situation at all. For massive magnetic data storage, tape is still very valid. You're just not going to find 500GB HDD's with such low failure rates in 10-packs for $1000 like you can get tapes at today. And tape can drop in price much more easily than HDD's will.

I'd give it a good 10-15 years before our massive tape storage units disappear from the datacenters.

Wait...

[mc moss]

I thought this was the year of the Linux desktop.

Re:Wait...

[montyzooooma]

It is, and it'll have a SSD.

Virtually there with the Eee PC anyway.

Re:Wait...

[Anonymous Coward]

Well, it looks like the year of many things:

2008: The year of the big airline merger
2008: The year of RSS
2008: The year of OpenID
2008: The year of layout engine - CSS3
2008: The year of principles
2008: The year of Palestine

And all along I thought it was the year of the rat...

Re:Wait...

[Loibisch]

You forgot that it is probably also "The Year of Linux on the Desktop" again...like every year before. :)

(sidenote: 2007 was "the year of Linux on MY Desktop" at least...+1 to the crowd)

Re:Wait...

[claytonjr]

You forgot the obligatory 2008: Year of the Linux Desktop.

Re:Wait...

[Anonymous Coward]

You forgot the obligatory 2008: Year of the Linux Desktop. You forgot to mention that that's already redundant again.

Re:Wait...

[Anonymous Coward]

It can be both! Linus Torvalds said one thing he's watching in 2008 is solid state drives.

Re:within 5 years, tape manufactuers will have tro

[JerryLove]

As I understand it, the problem with using HDDs for backup, at least archival backup, has more to do with longevity than anything else. An LTO tape has a shelf life of 30 years. HDDs don't.

Retrofitting existing equipment?

[Anonymous Coward]

Is there any hope of retrofitting old existing boxes with these drives or are they all going to come with newer (eg: SATA) interfaces?
I have a few older servers which will be too expensive to completely replace, but it would be nice to have solid state drives in

Re:Retrofitting existing equipment?

[thomasdz]

Is there any hope of retrofitting old existing boxes with these drives or are they all going to come with newer (eg: SATA) interfaces?
I have a few older servers which will be too expensive to completely replace, but it would be nice to have solid state drives in

Hasn't SATA been around for five years? I agree it would be nice to keep old interfaces around, but at some time, you're going to have to buckle up and replace your server if you want to use new technologies

Re:Retrofitting existing equipment?

[Anonymous Coward]

Hell, you can get addon SATA PCI cards. You have to have a _really_ old server for it to be totally incapable of SATA.

Re:Retrofitting existing equipment?

[somersault]

Alert generaaal!

Reports I Continue to Hear

[Nom du Keyboard]

Reports I continue to hear is of blocks going bad (meaning that overall storage is reduced by measurable chunks, rather than failing all at once the way a head-crash on rotating media can happen) in as short as weeks of use. Especially when the drive is rather full to start with, since wear leveling doesn't tend to move stored data to empty slots.

Until that time is years, instead of weeks, I don't see myself preferring more expensive, or even equal cost SSD, over rotating media drives.

Re:Reports I Continue to Hear

[Anonymous Coward]

HDD have the same issue they deal with it by including hidden storage space that only get's used as sectors goes bad. However, this increases random seek time on otherwise sequential data which is a PITA.

PS: A SSD with wear leveling last far longer than HDD when you do lot's of random R/W.

Re:Reports I Continue to Hear

[BitZtream]

Theres no need to move data already stored, the damage is caused by writes to flash, not reads.

Re:Reports I Continue to Hear

[lazyforker]

Just curious (and I suspect some other readers are too) - do you have some references/reports/links I could read? If an affordable 2.5" SATA SSD comes out soon I'd seriously consider replacing my Mac Mini's HDD ASAP. But not if the wear levelling is ineffective!

Re:Reports I Continue to Hear

[RicktheBrick]

Can anyone believe what is coming out of CES? Anyone can get a booth there just by paying the fee. It does not matter if the technology even exist. Atom Chip has been there several times. How many people really believe that their technology exist? Just do a search for Atom Chip here on slashdot and see the amazing claims they have been putting out for the last three years(6.8GHz laptop, 1TByte solid state memory chip). They were awarded even this year with the honoree award. Anyone can get that by paying for it. Calvin and Hobbes could get it for their cardboard box which transported them back into the age of dinosaurs if they paid the fee.

Re:Reports I Continue to Hear

[Just Some Guy]

Especially when the drive is rather full to start with, since wear leveling doesn't tend to move stored data to empty slots.

Asking for about the 10th time: does anyone know where that bad block map is stored? If "in another flash block", then isn't there a real (and relatively low) limit on how often you can remap blocks?

Re:Reports I Continue to Hear

[Anonymous Coward]

Reports I continue to hear is of blocks going bad (meaning that overall storage is reduced by measurable chunks, rather than failing all at once the way a head-crash on rotating media can happen) in as short as weeks of use. Especially when the drive is rather full to start with, since wear leveling doesn't tend to move stored data to empty slots.
Would you cite that? It's contrary to what I was taught about wear-leveling algorithms.

Re:Reports I Continue to Hear

[mako1138]

It's likely implementation dependent. In the Flash I've used, the first block has guaranteed 100K or so write cycles, so you generally stick the bad block map there. For a SSD, I'm sure things get very complex.

Re:within 5 years, tape manufactuers will have tro

[erroneus]

...and I have known at least two women who seem to enjoy being tied up in magnetic tape. Not only is it 'good memory' but it seems to have an effect on my hard drive systems as well.

Obligatory Wear Leveling Remark

[Anonymous Coward]

Allow me:

Initial Poster: These drives will not last! Solid state drives have limited read/write cycles.
Response: Dude! You are such a n00b. Newer versions of SSD have wear leveling to correct this problem!
Another responder: I am getting a kick out of these replies. We service 10,000 SSDs at our company and they only last 4 months.
Response: You suck!

Re:within 5 years, tape manufactuers will have tro

[jaweekes]

I actually see the solid state drives replacing tape (if the cost goes down). They would be smaller then tape, and the life-span would be as good. You could technically have a SSD loader, working like the current tape loaders, and the only thing needed would be a good connector that can take several thousand insertions (like the SD connector). SSD would be more reliable then tape because of the reduced mechanical parts.

SSD's would have all the advantages of tape (portable, easy to load, etc) without the mechanical problems that tape has. Wow, I need to patent this now!

It's not just for laptops...

[johnmcd]

Looks like the big boys are getting into the game also: EMC in Major Storage Performance Breakthrough; First with Enterprise-Ready Solid State Flash Drive Technology Market-leading Symmetrix DMX Systems to Feature Newest Flash-based Technology for Unprecedented Performance and Energy Efficiency http://www.emc.com/about/news/press/us/2008/011408-1.htm They're claiming a 10X performance improvement, but at 30X the cost/MB. Given that a high-end DMX holds around 3000 drives, that a lot of flash memory! John

Re:It's not just for laptops...

[BrianHursey]

Looks like you beat me to the posting.. I saw this this morning in my email. We will see how many customer bite into it.

Re:It's not just for laptops...

[Anonymous Coward]

The beauty of an enterprise storage system like EMCs Symmetrix is that, coupled with Documentum content management, you can set up automated lifecycle management and service level assignment so that the content that most needs to reside on your insanely expensive SSD storage at any given time is the content that is written there, and if that content is later reassigned to a lower service level (permanently or temporarily) it is automatically moved to more appropriate (read: cheaper) media. Done properly, this happens on-the-fly and automatically.

So you don't need to buy 100TB of flash storage for your 100TB of enterprise content ... you just need to spend a little time thinking about the value of your data and planning your storage infrastructure.

Re:It's not just for laptops...

[Apparition-X]

Just to clarify: the performance expectation is 30x the number of IOPS than an equivalent capacity hard drive, 1/10th the response time.

Re:It's not just for laptops...

[Anonymous Coward]

Maybe EMC will also announce that they are bringing back DG/UX. Maybe I'm just an old timer, but god I miss it.

Re:It's not just for laptops...

[BrianHursey]

haha, that was a good operating system.. I work with the people who support it and I think they are supporting it and aviion systems tell the end of 2008. http://www-csc.dg.com/csc/aviion_customer_letter.pdf

Stupid tech, making me feel old

[Sockatume]

When I was a lad, just starting out at university, back when nobody uncool knew what an iPod was, the biggest you could get was something like a 15GB version for £250, and it was a hard drive because as a rule of thumb, you couldn't fit more anything of value on a flash player. I was browsing on play.com last night, and Creative are doing a 16GB all-Flash player for about £160. That in itself wasn't surprising, as 16GB isn't much these days, but thinking back, this was one memory technology completely replacing another in a particular application. Really, I could see SSDs completely replacing hard drives in certain consumer products at the rate this is going. Joe and Jane public's internet box doesn't really need more than 50GB, does it? It'll keep them storing 10-megapixel holiday snaps for years.

Re:Stupid tech, making me feel old

[mini me]

I've somehow managed to accumulate 50GB worth of data and I don't store anything of interest. If I was storing photos and video like Joe Average does I'd imagine I'd be looking for a new hard drive right about now.

On the MP3 player side of things, realistically, you can't even listen to 1GB worth of music in a day. Because it's synced with your computer it doesn't matter if you need to remove some songs to make room for more. Therefore if you synced your MP3 player daily you wouldn't need more than 1GB of storage (assuming you want to listen to new music each day). On the other hand, I wouldn't want to remove photos on my computer to make way for newer photos. That's why you need a huge drive in your computer, but you don't need much space at all in your MP3 player.

Re:Stupid tech, making me feel old

[jackpot777]

Joe and Jane public's internet box doesn't really need more than 50GB, does it? It'll keep them storing 10-megapixel holiday snaps for years.

Video files can be quite big, but I see web-based storage handling that in the coming years too. Even if you're taking video at 720p (1GB per 30 minutes of home movie), it'll be relatively simple for someone with basic computer skills to throw the clips together into one home movie and upload them to some future YouTube HD (or competitor) site. Or it should be: it's bloody simple now. And if it's just a stupid home movie, the 320 by 240 resolution of YouTube is good enough right now for the "here's my holiday to Cancun / Ibiza / Blackpool" crowd, even when enlarged to full-screen. Upload your holiday vid, delete the larger file (or back-up to a CD or DVD ROM), easy.

Re:Stupid tech, making me feel old

[jackpot777]

...but it IS nice to have so much stuff on your iPod that it throws the occasional shuffled song at you that makes you think "what the...?"

I had it two minutes ago. I don't remember putting "Move" by Moby into iTunes. But there it was. Nice. Really nice.

Re:Stupid tech, making me feel old

[FunkSoulBrother]

I used to stress out about this too, as I really enjoy the surprises. I've found the key is to write a script that takes MEDIA_PLAYER_HDD_SIZE of random files from your PC and writes them to your iPod while docked overnight. That way you can enjoy randomizing your huge 100GB collection and still fit it on a non-bulky MP3 player which has plenty of room for 1 days' music.

Re:Stupid tech, making me feel old

[mini me]

That is why I have iTunes automatically put random songs on my iPod every time it syncs. This achieves the same effect without the need for a gigantic drive.

Re:Stupid tech, making me feel old

[u38cg]

Yep. The personal media player market has the most to gain from the move to SSD - broken units are what made things like the Cowon X5 uncommercial. There's a reason it's suddenly become very hard to find a 40GB+ music player that doesn't suck ass. These units are what is driving the SSD market at the moment - once capacities of 64-128GB become affordable they will then start to take off in laptops, and finally on desktops.

Oh really?

[Anonymous Coward]

I was sure it's the year of the Linux desktop...

EMC Planned Offering

[artgeeq]

Here is the article, though it is limited to the high-end Symmetrix. http://www.infoworld.com/article/08/01/14/EMC-will-replace-disks-with-solid-state-drives_1.html I wonder if we could do the same thing with a few dozen thumb drives.

Re:EMC Planned Offering

[Anonymous Coward]

Sure, you can already do it with ZFS. For fun: make yourself a zraid of thumb drives, put data on it, pull all of the drives plug them back in in random order, and watch ZFS sort it out.

Say no to moving parts

[eebra82]

We are finally starting to move away from a long era of computers with moving parts. Since conventional hard drives will be gone within 10 years (my prediction), all that remains is the media player (CD, DVD, etc). Obviously, I am not taking fans into consideration since I don't consider it to be a part of a computer system like a processor is.

Hopefully computers will be completely free from moving parts in 10 years or so. Now that would make it interesting for laptop owners.

Re:Say no to moving parts

[doti]

CDs and DVDs are already more dead the the hard disk.

Re:Say no to moving parts

[BrianHursey]

One Laptop Per child has no moving parts. Sooo you don't have to wait for 10 years. http://laptop.org/en/laptop/hardware/specs.shtml

Re:Say no to moving parts

[Thanshin]

Fans will be replaced by water heaters and geeks will have to shower at least four times a day to keep the computer cool.

Re:Say no to moving parts

[_KiTA_]

We are finally starting to move away from a long era of computers with moving parts. Since conventional hard drives will be gone within 10 years (my prediction), all that remains is the media player (CD, DVD, etc). Obviously, I am not taking fans into consideration since I don't consider it to be a part of a computer system like a processor is.

Hopefully computers will be completely free from moving parts in 10 years or so. Now that would make it interesting for laptop owners.


And CD/DVDs can easily be replaced with HVD Cards.

Re:Say no to moving parts

[eebra82]

Apparently, I didn't make myself clear. I'm not saying there is no alternative to CD/DVD/etc. Obviously, there is, with the most prominent being a simple internet connection. Eventually, everything will be downloaded instead of mounted via a disk. My point is that you don't buy a computer without a media player today because the vast majority still require it.

Re:Say no to moving parts

[mini me]

all that remains is the media player (CD, DVD, etc)

The network replaced the CD/DVD media players years ago. I would like to say that Apple will discontinue the CD/DVD drive in certain models as soon as tomorrow (like they did with the floppy in the iMac), but I'm not sure we have a decent alternative for OS installation yet, so that's going to keep them around for a while longer.

Re:Say no to moving parts

[AnyoneEB]

If more computers have the Linux-on-the-BIOS thing with HDD (SSD?) support, then that could probably be used to bootstrap OS installation using an internet connection. You could install Debian that way using debootstrap and I am sure it would be relatively easy to simply write out an installer to the main disk and then reboot into it. After all, that it what the Windows XP installer appears to do already except off a CD instead of the internet. I do not know how OS X installs work, but at least Microsoft does not care if you have a copy of their OS as long as they can require you to pay for a license. In fact, with an internet based install, it would probably be easier to require real activation. (I am sure it would still be cracked, of course, but it would require a bit more work.)

Re:within 5 years, tape manufactuers will have tro

[Albanach]

Actually you can pick up 10 500GB SATA Seagate drives with 5 year warranties for $1300.

The price per GB will continue to fall, so magnetic storage will be more cost effective. Of course there are other advantages and disadvantages to both.

EMC Solid Storage Array just anounced.

[BrianHursey]

EMC just announced this this morning. They are going to start haveing soldi state drives in there DMX-4 storage arrays. "EMC plans to offer flash drives in 73 GB and 146 GB capacities for the Symmetrix DMX-4 platform beginning later in Q1 2008". http://www.pcworld.com/businesscenter/article/141323/emc_readies_solidstate_drives_to_replace_disk_storage.html

SSD as a boot drive

[supertux]

The article talks about large solid state drives, but because of the price premium, I've been experimenting with smaller SSDs. In particular, I've been using an 8GB 266x CF card coupled with a CF->SATA adapter as the OS drive for my mythtv system for 5 months now with great success.

Not only is the flash drive completely silent, it is reasonably fast. Reads always benchmark at 40MB a second and writes benchmark at 34MB a second.

I've been a bit worried about the flash wearing out after repeated writes, but so far so good. Since my mythtv mysql installation is stored on it, as well as the normal system log files, I'm sure it sees quite a lot of action.

But to my point...

One common problem with systems such as mythtv that are under heavy IO stress is that during these moments of stress (lots of recordings going on at once) the whole operating system grinds to a halt or at least becomes sluggish waiting on some needed IO.

It was very common on my old mythtv setup where I used the extra space on the OS hard drives as extra storage space for mythtv recordings. I'm not experiencing any of that sluggishness with the new setup.

This has got me thinking that for my future desktop system, maybe instead of getting a raptor for the OS drive, and a large, slower hard drive for the rest of my stuff in order to minimize IO bottlenecks, I should swap out the raptor for a 16GB SSD for the OS drive. I'd end up with something that has almost no latency, good speed, silent, and it may be possibly just as reliable in that role.

What do you think?

Re:SSD as a boot drive

[vrmlguy]

I'd use unionfs to mount a tmpfs on top of your SSD for log files and such. Have cron run snapmerge periodically to copy the changes back to the SSD. This will let your SSD enjoy a long and useful life. See http://www.linuxjournal.com/article/7714#N0xa50890.0xdfb370 for more info.

Re:SSD as a boot drive

[DamonHD]

This is what I already do for my primary Net-facing server. Boot and most-frequently-accessed stuff on a 4GB SD card, the rest on the HDD which sleeps most of the time.

http://www.earth.org.uk/low-power-laptop.html

Rgds

Damon

Re:SSD as a boot drive

[Ash-Fox]

This is what I already do for my primary Net-facing server. Boot and most-frequently-accessed stuff on a 4GB SD card, the rest on the HDD which sleeps most of the time.

http://www.earth.org.uk/low-power-laptop.html

Heh, I've been thinking about doing something similar with the Eee PC. I think the Eee PC pretty useful to run some small services for me, for the fact that it's low powered, comes with a built in UPS (battery), built in microphone (good for generating entropy), flash based drive (less prone to failure, less energy requirements), smaller and since everything is self contained (keyboard, screen) - I don't need a lot of physical space for it either.

Re:SSD as a boot drive

[Heddahenrik]

One thing that has been tested successfully in various places is to RAID (RAID0) the SSD-disks. It makes them about as twice as fast, and it should be possible to RAID them in bigger arrays too. As there is so little risk of one disk breaking down, there is no excuse to not RAID-0 them.

These disks still have a problem with speed on random write though. It's nothing for read-write databases where NCQ (SATA2) disks are faster.

Re:within 5 years, tape manufactuers will have tro

[Wicko]

Really? I would have thought that magnetic tapes were more prone to damage/errors. I have no experience with them, but I guess I automatically related them to VHS tapes or cassettes, both of which show signs of noise after being used X amount of times. That and the fact that a magnet next to it can really screw it up. Then again, same goes for HDD...

Re:within 5 years, tape manufactuers will have tro

[WindBourne]

In the end, ppl will pick what is convenient Back-up is going to be moved to being a service. No doubt BIG business will continue to run tape drives. But the homes, and small businesses will move to service approach instead. And most are using hard drives with compression.

CDMA works for hard drives too!

[Hasmanean]

As far as I know, hard drives encode data on the disk using simple binary waveforms. Communication systems are designed to use elaborate modulation schemes, and employ digital coding methods which make much more efficient use of the communication channel.

Hard drive makers could do something similar, like spreading the data over a number of physical bits on the disk (such as CDMA does.) Essentially, they would not be limited by the density of the data on the disk, but by the SNR (signal to noise ratio) of the magnetic medium, which I imagine is very high.

Taking this idea furthur, they could bifurcate their encoding methods into 2: a low latency one that retains the characteristics of existing drives, and a high-bandwidth-low-latency scheme which uses digital coding methods to spread each block of data over an entire cylinder for example, and has requires reading the whole cylinder to retrieve a single bit. This would be useful for storing video and large image data, which is retrieved linearly and usually buffered too, and does not require low-latency access the way normal files on a filesystem do.

Hybrid schemes are always better than simple implementations, if they provide a closer fit to reality.

Re:CDMA works for hard drives too!

[Hasmanean]

Sorry the second scheme I mentioned should read high-bandwidth/high-latency.
High-bw/low latency is what flash offers today, and is something hard disk makers can only dream about.

Re:CDMA works for hard drives too!

[maz2331]

Trust me, they are using some unbelievably complex modulation schemes to write the bits on the media. It isn't simple like BPSK or FSK.

Re:CDMA works for hard drives too!

[BillBrasky]

Hard disk drives defintely do NOT write simple binary waveforms to the disk. They use encodings specialized for magnetic media, such as Extended PRML. This is coupled with error correcting algorithms like Viterbi (same as CDMA).

Re:within 5 years, tape manufactuers will have tro

[blahplusplus]

"Tape has been used for eons for back-up, but I think that HDD will overtake that role as their prices will be forced to go way down."

There may be another revolution in hard drive technology if they keep pushing new head designs, i.e. "non-mechanical" heads (i.e. light/lasers/etc), how feasible and cost effective this is, is up in the air.

Re:within 5 years, tape manufactuers will have tro

[DeeQ]

Well, Im not sure about you but im my personal VHS and cassette players I never used a cleaning tape to maintain them. Using a cleaning tape on a monthly basis in tape drives seems to have great results. Also there is alot more ware and tare on a VHS or cassette player you are constantly pulling tapes in and out every couple of hours or so. With a tape drive backup its a daily type thing most likely which wouldn't put as much ware. Its all about maintaining them properly.

Re:within 5 years, tape manufactuers will have tro

[Apparition-X]

With respect to the "trust" issue: this is perception, not reality. Disk drive arrays of the type used for backup (virtual tape) have a reliability rating of 99.99% to 99.999% for uptime, and lose or corrupt data even less than that. The very best tape will approach 99.5% but only if it is not physically moved outside of a tape library--like it would, for example, it taken off site for disaster recovery purposes. The moment that humans start to handle it, bump it, get it dirty, etc, reliability drops below 99%, sometimes well below to the 95% level or so. So, very best case, disk is at least an order of magnitude more reliable than tape, and most often, several orders of magnitude.

full disclosure: I work for a company that sells virtual tape. Nevertheless, all the data is backed up by vendor neutral analysts like Gardner, and tape companies themselves.

Re:within 5 years, tape manufactuers will have tro

[Sen.NullProcPntr]

An LTO tape has a shelf life of 30 years. HDDs don't. You are probably right but are there any numbers on the expected shelf life (powered down) of a HDD? The typical 1-5 year warranty assumes normal usage with a certain number of power on/off cycles and some number of MBs written and read per day. What about the case of write once, power off for a long time, then read? It may turn out that HDDs are more durable than tape.

Or looking at it from the other side - how long would a tape last as your swap media?

Re:within 5 years, tape manufactuers will have tro

[Wicko]

True, and I guess VCRs and tape decks aren't really made with longevity in mind (instead, use cheap parts to sell to the consumer for lower prices), and could cause some of these problems with VHS or cassette, whereas the entire purpose of a tape drive is to backup information, and longevity is the main factor in a purchase..

OS drives.

[Tatarize]

Yeah, certainly I'd pay double for storage for a reasonable 20-30 gig SSD. The seek times and lack of instant crashes out are great. For the speed and reliability, I'd replace my harddrive with a downright tiny drive just to run my OS. Their clicky cousins can store my massive amounts of non-critical data. I'm fine using magnetic drives I just don't want to use them for data I access all the fricking time.

I think we'll start seeing the drop off of magnetic drives well preceding the the overtake. We'll see people buying cheap 4 TB drives a couple years from now, but as is most basic users can't fill the 40 gigs. Why do they need more than a few gigs for a few dollars? If they have storage needs... that'll be your MM's job.

Re:OS drives.

[vrmlguy]

We'll see people buying cheap 4 TB drives a couple years from now, but as is most basic users can't fill the 40 gigs. Why do they need more than a few gigs for a few dollars?

Are you saying that most users don't have access to porn newsgroups and/or ripped DVDs? LifeBits can be expected to generate a lot of data, as well.

Re:OS drives.

[Tatarize]

No. I'm saying that if they happen to have 3TB of porn and movies, mp3s and TV shows... they can put it on the slower cheaper drive that doesn't load up lots random files all the time.

Re:within 5 years, tape manufactuers will have tro

[SQLGuru]

I'm sure there are enough old hard drives floating around places that accept used computer equipment (Goodwill, Tech museums, collectors) that we should be able to do a quick survey.....provided someone can find a working RLL controller.

Layne

Flat panel/CRTs all over again

[JerryQ]

In (approx) 1992 I went to SID (society for information display) in Florida, and a keynote speaker said, roughly: "I have been coming here for 30 years, and I expect to hear, just like I heard 30 years ago and most years since, that within 10 years flat panels will overtake CRTs and make them redundant. Why has this not happened? Because CRT has continued to get cheaper and better quality, thus removing the opportunity for flat panel, because the goals keep moving" He also pointed out that we would get there (and we have) but that we should never underestimate where old technologies can go. In 1983 I put together a business plan for an outsourced proposal I was working on, and we put in £17K (thats $28k) to cover a 70 megabyte hard drive. Now I see one inch drives in iPods carrying multi gigs. I believe that we will see phased take up, ie where it is needed most (e.g. like the way airlines put in flat panels instead of CRTs to reduce weight), before the HDD manufacturers will curl up and leave the scene. Jerry

Re:Flat panel/CRTs all over again

[xtracto]

I was working on, and we put in £17K (thats $28k) ,
Haha, more like $33,449.08 without trading commission (according to XE).

I believe that we will see phased take up, ie where it is needed most (e.g. like the way airlines put in flat panels instead of CRTs to reduce weight), before the HDD manufacturers will curl up and leave the scene.

We are already seeing this. SSD have opened the possibility of new devices. More than replace the technology from old uses it will allow new uses. I have always thought that, if you can have 4GB in a very very small micro-SD drive then, ho much memory can be contained in the space of a hard disk if you use the same technology!.

more like..

[pak9rabid]

2008, the year of predictions

all prices are falling

[peter303]

Digital video tape is approaching a tenth of a cent per gigabyte. Still going to be one to three orders of magnitude lower than flash. Moving disk will be in between.

What about limited write cycles?

[unassimilatible]

And, of course, any technology with no moving parts is preferable-- mechanical parts have an annoying tendency to freeze up with vacuum thermal cycling.

How are the SSD makers getting around the limited write cycles of flash drives? Flash, even high endurance can actually wear out faster than HDDs with all of Windows' endless writing to the page file. A fluid bearing HDD can last a long time theoretically. One of the problems with those few people who have managed to get Windows (XP, not PE) to boot from a flash drive is that Windows will burn out a flash drive pretty quick with its endless page writes. Yes, you could turn off the page file, but that isn't how XP is supposed to work. Then again, neither is BSOD.

I understand that Linux pages too, but I'd imagine the average Linux nerd would implement a workaround.

Re:What about limited write cycles?

[Wolvie MkM]

Here>/a>, there's a mention about file systems that will accomadate the problem that SSD drives have. Basically spreading out the bits all over the drive if my interpretation is correct.

As for XP, you don't "need" a page file if you've got over a Gig and don't do anything that will come close to the limit. I've removed the page file on all of my systems and been fine (3 Gig Desktop, 2Gig Desktop, & 1Gig notebook). It's an easy way to save some space if you know your limits!

Cheers hope that helped

Re:What about limited write cycles?

[Wolvie MkM]

Apologies for apparently not knowing how to work the HTML tags or spell, Jesus this morning sucks lol...

Re:What about limited write cycles?

[v1]

An intelligent flash controller could level the write field by using a table to look up blocks, and periodically shift the table offset so that writes were being distributed more evenly across the flash storage. Doesn't help much until you consider the very small percentage of space the page file occupies. It would take a long time for it to wear out the entire drive.

But this would do scarry things to the complexity of the controller chip. Though from what I've seen of the ZFS specs, it should be a piece 'o cake

Believe this nonsense not

[Anonymous Coward]

Nothing wrecks the future better than predicting it.

It's very much safer to predict the past, less ways to go wrong.

There is the Fry's Factor

[bigredradio]

SATA was around for a long time, but it wasn't until I noticed that you could not order a system with IDE, or buy a hard drive at fry's that wasn't SATA that the transition really took off.

I suspect that same will be with SSDs. Consumer demand will bring down prices, but the true test is whether it becomes main stream or a replacement for SATA. The best technology is not what always gets adopted. Sometimes adoption has more to do with marketing.

When I see it at Fry's overtaking traditional disks, I will know it has happened. Before then, I wouldn't count on it.

Re:within 5 years, tape manufactuers will have tro

[Sen.NullProcPntr]

I'm sure there are enough old hard drives floating around places that accept used computer equipment (Goodwill, Tech museums, collectors) that we should be able to do a quick survey... That would be a good experiment, but it's not exactly the same thing - all those drives would have gone through the whole power cycle, read/write lifetime already. What I'm talking about is a HDD that gets very little use - write once, wait many years (in a controlled environment), read once.

..provided someone can find a working RLL controller. Another good point - not just for HDDs but will today's tape drive read a 30 year old tape? Do tape _drives_ last 30 years?

Re:Reports I Continue to HearWEAR-LEVELING DONE RI

[Nom du Keyboard]

Theres no need to move data already stored, the damage is caused by writes to flash, not reads.

The point of moving existing data is that the damage is done by writes. And long-term data has no writes performed to it. Therefore, for wear-leveling purposes, you would desirably wish to move the long-term data to a heavily written, but not yet failed, area of the flash, since it wouldn't be written again, freeing up the seldom-written, or even once-written, area for more use.

Really?

[Anonymous Coward]

I'm not sure we have a decent alternative for OS installation yet, so [...] Really?

Re:Really?

[mini me]

Find me a DVD-sized USB drive that can be distributed for the same price as a CD/DVD and I'll agree with you. This is an economics issue, not a technological one.

Re:within 5 years, tape manufactuers will have tro

[Anonymous Coward]

Hmm...a hard drive that uses a laser as a read/write head...interesting. I wonder what they will call it? Compact Disc, perhaps? Digital Versatile (or Video) Disc, maybe? Oooh, I know let's use a blue laser and call it Blu-Ray or maybe High Definition DVD.

How long before..?

[Lord Bitman]

How long before buffering reads into memory is considered archaic legacy support and something to be avoided?

Ask Slashdot

[pragma_x]

SSD is expensive right now. Is there any kind of DIY solution for battery-backed RAM out there? How about hacking one together?

Re:Ask Slashdot

[hibji]

Are you talking about something like this?

http://en.wikipedia.org/wiki/I-RAM

RAM is still more expensive than flash though right?

Re:Ask Slashdot

[32771]

Well I found that you could get 1GB DDR-2 RAM at 15 euros. Interestingly SDRAM and DDR RAM will be more expensive per GB if you buy them new. If you look for 1GB Flash you can get a CF card for 12 euros. Your only way to grow your own SSD is if you can get used RAM much cheaper than 12 euros.

The next question is how do you want to hook it up to your machine, certainly through PIDE. I suppose a Spartan3 could do the job since it can also talk to DDR2-RAM.

Then comes the interesting part. How much is your time worth?

You have to come up with a design and layout for your PCB. You will probably get a multilayer PCB which will be costly. If you are lucky you can populate the board yourself, especially if you can go without BGAs (unlikely).

Then you have to test the result and program the FPGA. If you messed up your PCB layout you are back to getting a new board and doing some more soldering and testing.

I have not done any design with IDE interfaces so I can't say how difficult this will be (it is probably the simplest). Also I don't know how to connect the DDR2-RAM to the Spartan.

This will take you some months (4-6, you do have a job do you?) by then the FLASH memory price will have fallen further.

My Samsung SSD 32GB has cost me ~320 euros. You will have to get below that with your design supporting the same amount of memory.

160 euros for 32GB (if you are lucky and get it for half the price)
  30 euros for 4 prototype PCBs (my estimate)
  60 euros for one larger spartan (XC3S1500-4FG456C)
  50 euros for a JTAG adapter (if you don't have one)
    0 for Xilinx tools
    0 for your time
  -? for the learning exercise
    0 for the lab you already have (should include a logic analyzer and a decent scope otherwise add cost for renting one)
-------------
300 euros sum

Uh right, add a decent Li-Ion battery and add 20 euros.

I guess you have figured out by now, that waiting for 6 more months twiddling your thumbs might get you there cheaper.

Re:Ask Slashdot

[pragma_x]

I guess you have figured out by now, that waiting for 6 more months twiddling your thumbs might get you there cheaper.

I had a feeling it was going to be like that, but I figured: hey, I'm not the smartest guy around here, so let's ask! F or $420 USD (conversion rate of 1.4 dollars per Euro - I think), I could do better with a Compact-Flash IDE adapter and some nice CF chips and still have some cash left over. Granted, the performance wouldn't be quite the same, but it would work.

I know that Gigabyte has/had a PCI based solution for battery-backed RAM that started at about $140 USD (w/o RAM). I was hoping that my Google-fu was simply weak and that I simply wasn't seeing the competition in this niche. Now it looks like SSD came along in time to kill it.

Thanks for taking the time to write such a thoughtful reply.

Re:Ask Slashdot

[pragma_x]

Yea, that's one such product. I held out hope* that with SSD's hitting the consumer level soon, that we'd have a wider range of products to pick from within that niche. I guess we're not there yet since (like you say) RAM isn't cheap enough to edge out flash.

(* Along with the incredibly vain hope that this would be "the year of open hardware" that might make recycling free-schwag thumbdrives a worthwhile prospect)

Re:Ask Slashdot

[32771]

I liked the challenge of the DIY part of the question. There are occasions where I would ask the same. In most cases I end up with the same conclusion as with your problem, if the learning experience isn't worth it forget it.

But as I see at work it takes a nice lab and time to get anything interesting done. I rather try to learn something at work especially since your employer pays for it (more or less voluntarily) and you have far better means there.

What I sometimes think is that one could spend more time bastardizing things, i.e. taking things apart and making something new out of it - it still takes a lab though. Or maybe start a free software project if you have time and little money.

Actually I can't top the Samsung with an array of CF modules, USB sticks might work, but I don't wanna go there.

You could still try to find a cheap, used server mainboard which supports more than 4GB of RAM and connect it to your machine through GigE/NFS.
Uh, an Intel server mainboard - S3000AH - supports 8GB, costs like $120 on ebay, that could do, you can actually get a cheaper motherboard, something AMD based. Isn't this totally funny, you can actually, with some luck, find a cheaper mother board than that Gigabyte RAM disk :), and it even supports more memory, but it probably is slower (although there are inexpensive Infiniband adapters out there).

Solution for /var activity

[r_jensen11]

I've been thinking about this workaround for quite a while, but I never see anything discussed about it. But why not mitigate activity on the SSD by having /var be a ramdrive? Once the system is stable, cron a backup (snapshot) to the SSD, along with writes for whenever the partition is unmounted. Ram is cheap enough that for typical applications, /var shouldn't be too large (unless you have large caches stored in /var, but that can be solved with symbolic links)

2008????

[kellyb9]

Dammit, I was going to declare 2008 the year of desktop Linux... guess I'll have to wait until 2009.

Re:within 5 years, tape manufactuers will have tro

[afidel]

So what, I can get 10 LTO3 tapes (600GB average for our data) for $500. If we upgraded our drives to LTO4 I could get 10 LTO4 tapes (1.2TB average for our data) for $1300. Unless you use less than a handful of tapes a month tape beats HDD. Heck due to ongoing litigation I haven't recycled a tape since October 2006, using HDD for that would have cost an absolute fortune. There are markets for most technologies (otherwise why invent them) but I don't think HDD based backups are going to beat tape for many businesses anytime soon.

BitMICRO 832GB 2.5-inch & 1.6TB 3.5-inch SSD

[OtherFarm]

832GB 2.5-inch SSD
1.6TB 3.5-inch SSD

source - ofpblog

Re:within 5 years, tape manufactuers will have tro

[jabuzz]

Not only that when the tape is unloaded and sitting there in the library it consumes *ZERO* power. A couple of tape frames and you can easily exceed 1PB of storage for a fraction the price of hard drives and a vanishingly small percentage of the power. A good HSM later and you are cooking.

Not only that, what it would cost to reliably cable up all those drives I shudder to think. The nice man from IBM comes tomorrow to install another tape frame onto our library at work. We are then going to fit some LTO4 drives, load a bunch of tapes, and start shuffling it all off the LTO1 tapes currently in there. Once operation has been completed (there is a lot of data to move) and we remove all the LTO1 tapes we should have some 1.5PB of storage. Note there are of course two libraries for redundancy and disaster recovery.

Do that in spinning HDD, dream on mate.

Today

[SillyNickName]

Where's the news in this? I'll sell you one today if you want to pay for it.

Re:within 5 years, tape manufactuers will have tro

[ichigo 2.0]

Wow, I need to patent this now!

Your solution sounds like a powerful version of a memory card reader. :)

OTOH, you probably would be able to patent it. Sigh.

Re:within 5 years, tape manufactuers will have tro

[slaingod]

In one year DVD and Blu-Ray burn media manufacturers will be having trouble. Once 1TB hard drives reach $100 dollars, you are basically at break even media cost wise with DVD+R media. Meaning you can just store all of your downloaded movies on a harddrive at the same cost as DVD media. In 2 years-ish, hard drives will be the clear winner, with 2TB drives approaching that $100 range possibly. While you have the chance of catastrpohic failure, if you really are using these things just for occasional read storage along with a docking solution you end up saving time and money and physical space (those 500 sleeve DVD holders add $0.10 to every disk you burn).

It will be nice in say 5 years when SSD tech reaches that same threshold, since catastrophic failure of those drives is presumably much less likely to occur.

ZFS on flash?

[Anonymous Coward]

And even the 800-1000USD SSD drives fall way short of their predicted write lifetime when put into any environment where i/o is primarily small random r/w operations vs large sequential stream r/w operations. Write leveling is generally limited to 512b or larger blocks which makes it ineffective for these types of loads, furthermore the cheaper devices tend to use MLC flash nand vs. the less dense SLC.

What about copy-on-write file systems like ZFS that don't touch the data already on-disk, but only write to unused blocks?

ZFS is a bit of a more a general FS, unlike JFFS which tends to be in the niche of the embedded space.

Re:ZFS on flash?

[dfn_deux]

I've tested that use case as well. I can't profess to understand the exact method by which the two interact, but I experienced similar results with both zfs and ufs.

HDDs Are Dead, AFAIC

[gacl]

I bought a 16 GB Samsung SSD a month ago ( Christmas present to myself ) which was about 200 USD. It is now installed on my Dell Latitude X1 which now has no moving parts at all, and the computer is faster and silent. With 2 000 000 writes, i don't think i should worry about this drive failing anytime soon. And if it does. . . i back up.

I've put the old 1.8 HDD in a case and now i'm using it as storage, but once it dies i don't think i'll ever buy an HDD again.

Gus

250GB!?

[barl0w2]

Don't you mean 256GB? I mean we do know our SSD here right?

Smart Talent was demoing 256GB at the show last week, and I just didn't get a photo of the demo system running it. But in saying that it'll be mainstream in X months or whatever is just a gamble. If we have another earthquake in Taiwan or one in Korea, RAM prices are going through the roof as all the factories are in Taiwan, and Samsung owns the flash memory market right now.

Aside from that, here are 145 photos from CES of stuff that I thought was worth shooting with my camera at the show: http://www.flickr.com/photos/barl0w/