Top Solid State Disks and TB Drives Reviewed

Lucas123 writes "Computerworld has reviewed six of the latest hard disk drives, including 32GB and 64GB solid state disks, a low-energy consumption 'green' drive and several terabyte-size drives. With the exception of capacity, the solid state disk drives appear to beat spinning disk in every category, from CPU utilization, energy consumption and read/writes. The Samsung SSD drive was the most impressive, with a read speed of 100MB/sec and write speed of 80 MB/sec, compared to an average 59MB/sec and 60MB/sec read/write speed for a traditional hard drive."

Comments

Longevity of NAND flash

[ASkGNet]

NAND flash deteriorates with use. When used in a high-I/O situations like hard drives, just how much time will it be able to work correctly? If I recall correctly, NAND blocks are guaranteed to the order of 100000 writes.

Re:Longevity of NAND flash

[peragrin]

Yes, new on disk topology mappings, and new tech give you roughly a million r/w and the mappings help to evenly distribute the load.

Re:Longevity of NAND flash

[Anonymous Coward]

And on modern flash devices, writes are automatically distributed over the entire filesystem so that particular areas don't get hammered and worn out. With a large flash device with a reasonable amount of free space, the time taken to reach 100,000 writes to a particular bit will be pretty long (as long or longer than a conventional hard drive's MTBF according to various sources I've read (unfortunately, I don't have the links to hand).

The article (or the manufacturer?) is misleading though - figures of 100MB/s read and 80MB/s write are quoted, but the drive is benchmarked at about 25ish...

Re:Longevity of NAND flash

[plague3106]

This is always claimed as the solution, "evening" writes. But I think the question about how long will the drive last is still relevent; all it takes is a mostly full disk, which has a high I/O load. Even with evening, it seems that at least part of the disk can fail before the rest of the disk.

Do traditional drives fail if the same sector is written to over and over again as well?

Re:Longevity of NAND flash

[goofy183]

Will this ever die? The write cycle counts in modern flash is in the millions now. Doing the math you very easily get 20+ years before write cycle wear is a concern: http://www.storagesearch.com/ssdmyths-endurance.html

How many heavily used spinning drives do you know that last even 10+ years?

Re:Longevity of NAND flash

[baywulf]

Actually the endurance on NAND has been going lower over the years as they switched to smaller cell geometry, larger capacity and MLC technology. Some are as low as 5000 cycle endurance. These MLC(multi-level cell) NAND tend also to be much slower than SLC(single-level cell) NAND. Most SLC NAND have around 50K or 100K endurance.

Re:Longevity of NAND flash

[Bassman59]

NAND flash deteriorates with use. When used in a high-I/O situations like hard drives, just how much time will it be able to work correctly? If I recall correctly, NAND blocks are guaranteed to the order of 100000 writes.

Do a web search for "flash wear leveling."

-a

Re:Longevity of NAND flash

[orclevegam]

Thank you for posting this. This is a really great link and is going in my reference documents bookmarks. I'd call for others to mod you up, but they already have.

Re:Longevity of NAND flash

[Khyber]

And this is why we're moving away from NAND, so get that damned term out of your head already! OUM/OVM is coming, uses a nearly identical manufacturing process (It's the same thing found in RW optical media, except you use electricity instead of a laser to change it's state) as CMOS does, and it has FAR more read/write cycles than anything NAND could have ever hoped to achieve, in the range of 10^8 as opposed to NAND 10^5-10^6

Re:Longevity of NAND flash

[ComputerSlicer23]

You ever actually done this? I work on embedded systems that use flash drives... Even with write levelling, we've had failures. It's lots of fun, when your 512MB flash isn't 512MB, and will suddenly lose ~41MB suddenly. As a work around, we've had to start partitioning with extra space left lying around at the end of a disk. This isn't even a heavy workload system.

Some friends of mine at another company that were using them in a I/O laden system that wanted to replace laptop drives to make the machinews lower power and more reliable can blow out a flash drive in about 4 weeks.

Kirby

Re:Longevity of NAND flash

[dunkers]

He makes the mistake of assuming a complete erase each cycle of full disk. In reality, that sort of runaway process would fill the disk and then start erasing small parts to make room for more data. If 1K of space is made each time the disk if full then that actual cycle would be 1 x 64GB plus 2m x 1K.

Plug /those/ figures in and it turns out the disk will be trash within 20 minutes

Re:Longevity of NAND flash

[Lumpy]

How many heavily used spinning drives do you know that last even 10+ years?

I have at least 15 of them doing that right now. my last employer changed out the SCSI array in a couple of powervaults in 2005 I picked the drives out of the trash and have been using them in a powervault I got off ebay for $25.00 the drives have been spinning for over 10 years now.

I have had only 1 drive fail out of the "untrustworthy" ones I got out of the trash.

SCSI U160 drives are incredibly robust, not like the crap they have made over the past 5 years. Yes they are only 32Gig each but they work and work for over 10 years.

Re:Longevity of NAND flash

[MyNymWasTaken]

You've never had spinning platter hard drives fail on you?

Re:Longevity of NAND flash

[theoverlay]

With 3-bit and even quadbit MLC NAND around the corner we should see faster controllers that will make these drives more attractive and larger. There are even some hybrid controllers that allow multiple nand types(mlc and slc) and even nor in the same application. One of these is Samsung's flex-OneNAND. A good site for more information is http://infiniteadmin.com/

Re:Longevity of NAND flash

[Anonymous Coward]

The math on that page is wrong because erasing data on NAND flash can only be done in blocks, which are relatively large, often several to hundreds of kB; if you want to reset one bit the entire block must be reset. IIRC, this means that in the very worst case you can destroy the flash in days, not years as that link claims.

Re:Longevity of NAND flash

[ComputerSlicer23]

Yes I have. However, I've never had one magically get smaller on me in such a way that fsck decides that your done fixing the filesystem. With SSD, YES, I've had exactly that happen to me.

In my life, I've lost a total of about 42Kb be completely unrecoverable with spinning media (yes, I mean that number literally). I use RAID extensively, I was the DBA/SA/Developer at a place that had ~10TB of disk online for 5 years. In all that time, 42KB is all I lost. Oh, that was in the off-line, tertiary backup of the production database (it was one of 5 copies that could be used as a starting point for recovery, we also had the redo logs for 5 days, each DB was a snapshot from one of the previous 5 days). It was stored on bleeding edge IDE drives put in a RAID 5 array. We used it as a cheap staging area before pushing the data over Firewire/USB to a removable drive that an officer of the company took home as part of the disaster recovery system (it had only the most recent DB and redo logs). The guy didn't RMA the hot spare, and we had two drives fail in about 3 days while the hot spare was waiting for the RMA paper work to fill out. In that one particular case, using ddrescue, I recovered all of the data off of the RAID5 array but 42KB (even though it was an ext3 filesystem on LVM, on a RAID5 array, which made the recovery even more complex). Every other bit and byte of data in my life from spinning media that I cared about, I've recovered (I've had a number of drives die with data I didn't care about, but I could have recovered from if need be). Trust me, I know about reliability, backups, and how to manage media to ensure that failure doesn't happen. I know about failure modes of drives. I've hot swapped my fair share of drives, and done the RMA paperwork. I've been in charge of drives that losing any one of the ~200 drives would have cost 10 times as much as I made in a year if I couldn't reproduce the data on it within hours.

If it had been worth $10K, I'd have sent off the drive to get that 42KB of data recovered. But it wasn't. It's well understood how the failure mode of spinning media. People know exactly how to do things like erase drives securely. People know who to call that has a clean room that can remove the magentic media to and put it under a microscope to get the data recovered. SSD isn't nearly as mature in that sense.

All of that is really to say: Yes, I know something about disks and drives. My point is to say that SSD's aren't magic pixie dust in terms of reliablabilty. I've had exactly what he's saying I shouldn't worry about happen to me on a regular basis. Enough, that our engineering department has developed specific procedures to deal with them in the field. We've changed our release procedures to accout for them. If your going to use an SSD or flash drive, go kick the crap out of it. Don't believe on faith anything you read on Slashdot (including this post, which is anecdotal). We order lots of 5,000 flash disk, and you can bet that at least 100 of them has serious flaws within being fielded. The ones the developers and testing uses regularly develop problems in terms of months, not years. The manufacturer tells us essentially, it's not worth it to find those, so deal with it.

The whole point of replacing the laptop drive was to make the silly thing more reliable. But making it uber-reliable for 4 weeks until the write leveling crapped out wasn't the idea.

Kirby

Re:Longevity of NAND flash

[zeet]

So you're saying that you lost 42kb of data you did care about, and some other unnamed amount of data that you lost but didn't care about? That seems a bit disingenuous. Even if you could have recovered the other data, since you didn't try it wasn't recovered.

Re:Longevity of NAND flash

[Amouth]

i know what you mean.. my desktop i am using right now has an IBM 36gb scsi drive that is pushing 9 years as we speak.. wonderful drives - they truly just don't make them like they used to ... on the other hand just for the sake of it (and it's age) i have a seagate 9.1gb scsi drive that takes up 3x5.1/4 bay's - it was one of the first 9gb drives on the market.. still running.. on a duel p pro running slackware.. it keeps right on chugging away and keep spam out of my mail box..

on the other hand i have 4 wd 250gb ide drives on my desk that give smart errors and just are damn flaky... what can i say.. you get what you pay for.

Re:Longevity of NAND flash

[Anonymous Coward]

Wrong. The drive controller keeps track of how many times a block has been erased and swaps it with a low-usage page when the count gets too high. This is how they do load leveling.

Re:Longevity of NAND flash

[TooMuchToDo]

I can understand your reluctance to trust flash media. Indeed, it hasn't been proven like spinning media has. Let's take another example. An in-car radio. I want a 100GB hard drive in my car, solid state, that is for all intents and purposes write once. I should be able to dump 10s of GBs of MP3s onto it, and the index should be stored on a replaceable CF card (as the index would be changed often). But why would I remove music from the drive? I can just add more music.

For the above example, a flash drive works very well. If you need the benefits of flash storage mediums (vs spinning media) you should be prepared to engineer around the situation. Run temporary data out of RAM with battery backup, and only commit the data to flash between reboots and power outages.

Re:Longevity of NAND flash

[vertinox]

Do traditional drives fail if the same sector is written to over and over again as well?

No, but they'll fail either reading or writing over time regardless if you are writing or just reading just because the drive is moving. Even if you cool your standard drive, eventually it could just fail because it was left on for 10 years (since an active drive is constantly spinning).

Now its not guaranteed to fail, but the chances of a standard HDD failing that you only read from and don't write it is far greater than a SSD that you put files on it one time and don't write further.

I think SSD shine in archival types of things that you don't plan on trashing and rewriting that often such as image collections, movies, and MP3s. That said, swap disks, scratch disk, and cache file directories would logically still have better performance on your spinning platter drives and if that drive goes belly up you haven't lost much.

Re:Longevity of NAND flash

[s_p_oneil]

"all it takes is a mostly full disk, which has a high I/O load"

It is a relevant question, but this wouldn't kill your hard drive, it would simply reduce the amount of free disk space. And it's not difficult to imagine a file system smart enough to move files around when this happens. When a sector gets written to too many times, it can simply look for and move a really old file onto that sector to free up some of the rarely used sectors of the drive. With the increased performance of SSD, you probably wouldn't even notice it.

Aside from the re-write issue, flash memory drives should be WAY more reliable than a mechanical HD. It should never just completely die or start getting bad sectors so fast you don't have time to retrieve your data. It should also be a lot easier to replace when it starts to degrade. It shouldn't be as susceptible to damage when you drop it from a height of 3-5 feet, or due to heat, cold, vibration, dust, humidity, etc. I'm not sure whether a magnetic field could erase it like a hard drive, but if not, that's another plus for SSD. I imagine SSD's are more susceptible to static electricity, but so is almost everything else plugged into your motherboard, so I'm not sure if that could be considered a minus.

I'm sure if you ever tried an SSD on a laptop, you'd never want to go back to an old HD. The improved performance and battery life would make going back to an old laptop HD seem like going from broadband back to an old 56K modem.

Re:Longevity of NAND flash

[Nom du Keyboard]

So you're saying that you lost 42kb of data you did care about, and some other unnamed amount of data that you lost but didn't care about? That seems a bit disingenuous. Even if you could have recovered the other data, since you didn't try it wasn't recovered.

I believe what Kirby was saying, in addition to SSD's crapping out in weeks instead of years, is that he can get the data back from rotating media virtually every time if it's important enough to be worth spending the $$$s on. Unimportant stuff he doesn't bother to spend the time and money on.

I believe he is also saying that "dead" rotating drives can still have their data recovered, while "dead" SSD drives cannot with current methods available.

As a user who had a lightly used Jump Drive die suddenly after 4 years, I can attest that the failure was complete, and every possible online recovery tool tried recovered nothing, as well as discouraging the idea of actually sending it in for full disassembly and attempted recovery. It was simply dead.

And this is not even bringing in the question of constant and sudden decreases in SSD drive capacity. How would you feel about a rather full regular hard drive that was suddenly several percent smaller? That could kill your system right there, even though most of the SSD was intact.

Re:Longevity of NAND flash

[Nom du Keyboard]

Do a web search for "flash wear leveling."

And you do a search to see how well that works when your SSD is mostly full, and the swap space is getting hit hard. Leveling doesn't tend to move static files often, meaning when the SSD is mostly full, only a small part of it is getting continually whacked. And when that goes out of service, you have an even smaller pool of free space to handle all the activity.

Re:Longevity of NAND flash

[BBandCMKRNL]

But why would I remove music from the drive? How about when you sell or trade in the car?

How about when your teenager plugs in their 50GB mp3 player and it gets auto-loaded to the car-audio system and your half full system is now full?

Re:Longevity of NAND flash

[psydeshow]

Just remember to mount these drives noatime to avoid a write every time you read a file.

For that matter, noatime is a sensible default for any desktop OS. When was the last time you actually searched for files you hadn't accessed in six months?

Re:Longevity of NAND flash

[TooMuchToDo]

How about when you sell or trade in the car?

You'd wipe the device once, leaving the drive empty for the next person (as well as hundreds of thousands of writes available). I said treat the device as being write once, not actually making it write once. You don't even have to wipe the drive if you're one of those types of non-copyright-obeying folks.

How about when your teenager plugs in their 50GB mp3 player and it gets auto-loaded to the car-audio system and your half full system is now full?

Remember on floppies the little plastic that you could move to enable write only? Think about that, but in software. If you hook up an iPod, the in-dash system should automatically play music from it. You should be able to go into a menu and copy from the iPod to the in-dash system, with it asking you to confirm before proceeding.

Also, if you're worried about running out of space, that's because we're only at 64GB flash drives. Wait until we get to a terabyte. I wish you luck filling a terabyte drive with music that you'll have time to listen to in this lifetime.

Any other concerns you can think of?

Re:Longevity of NAND flash

[Anonymous Coward]

Granted, your anecdote is a little extreme (I've had 20% of my platter drives fail on me, mostly due to my packing them up every 3 months and driving them for 6 hours between home and school), but what you say about the manufacturers' attitude towards SSD failures is paritcularly poignant.

It's not about the rated maximum writes or life cycle. The number on the packaging means nothing if even a small percentage of the physical product won't even come close to lasting that long. Numerous flash-based media have failed on me within two years despite the assurances otherwise on the packaging, and this is with a light write cycle (new data about once a day). The manufacturers' attitude explains why two of the same product bought at the same time could exhibit the vast differences in lifetime I've witnessed.

Re:Longevity of NAND flash

[cheater512]

SSD doesnt shine at archive.
Archive does not need fast speed nor good seek times.

Normal hard drives have plenty of speed for archive however, would be spun down most of the time (no wear and tear) and they provide what SSD cannot: capacity.
Having 64gig of data archived is great and all but at home I have 900gig of archive data. Small problem dont you think?

Re:Longevity of NAND flash

[ComputerSlicer23]

When I was young and stupid about drives and media, I lost a 1.2GB WD drive and lost everything on it. I couldn't spell "mkfs" or "fsck" and had no idea how to recover the drive at the time (I also didn't have the money to have a second drive to recover too, and no credit card so I could hold onto the first while having the second during the RMA). I was just young and ignorant. I lost a 1-2GB laptop drive that I literally just rode into the ground, I could have copied everything off and moved along. I knew the drive was going bad, but it was just a knock around system that I didn't care about. In the end, had I been thinking, I'd have saved the e-mail on it. I lost the first ~5-6 years of e-mail I had, but who wants e-mail from when they were 18-24? That was probably a couple of hundred MB that I might regret, but of nothing more then sentimental value. I'd never read it, and only be amused that I could prove I'm getting the same chain letters 15 years later.

I believe I had 4-5 drives I lost due to a virus or pilot error, but not a mechanical/media problem.

I've RMA'ed probably 100-200 drives due to some type of failure. I've had lots of of drives fail that were in a RAID array, that the mirror saved me. I've had lots of drives fail that were stand alone that had a section of bad sectors. All of that I recovered every byte of data from. Normally a drive that is going bad, you can still recover from for a very limited amount of time. Normally you have plenty of lead time, especially with SMART drive monitoring that your drive is going south. As long as you pay attention, spinning media isn't that hard to keep in good shape.

As a professional IT person, 42KB is it. On machines where production work is done for money at a company. 42KB is it, and in that case I was bound and determined to recover absolutely everything, and I invested a week into that project. I gave up on the 42KB once I proved that it was in a backup for the database that was at that point 15 days old (and thus of no use). Had it been necessary or cost effective, I'd have spent the $1-3K to get that drive images recovered by a professional data recovery shop. I think I've lost a drive or two on my personal machines at work, but the drive was fine, the laptop SATA controller was overheating. Using FSCK, I recovered the entire FS once the RAID controller was replaced. I think I had to re-rip some music from CD, because I failed to back it up prior to sending the laptop in for repair. I re-imaged the drive just to be safe in case the RAID controller had corrupted something important on the OS drive, which was the only reason I actually lost the music.

Again, it's the fact that the flash drives we have decided the drives are smaller at the interface level. Using fsck just scragged the system pretty much start to finish. I don't have a clue where the missing blocks are from. I have no idea what happened, upon reboot it decided that the block devices was smaller. Filesystem recover tools haven't had a chance to mature to understand those types of failures. Flash makers haven't yet decided that access to diagnostics and re-mapping logs might be of value to data recovery tools (at least none that I'm aware of). Access to the raw data (in case they are holding blocks in reserve). All of these things are reasons to be concerned about write leveling.

Kirby

Re:Longevity of NAND flash

[operagost]

They didn't have U160 SCSI ten years ago, and the standard sizes for SCSI disks don't include 32 GB (I assume you mean UW SCSI at 40 MB/s and 36 GB).

Re:Longevity of NAND flash

[ComputerSlicer23]

I think that makes perfect sense, but then I'd think that all the money they'd spent in making the thing perform faster then say 1MB/sec read rate is totally wasted. I'd assume folks are trying to push these are replacements for enterprise server machines, which I'd be extremely relucant to do.

Folks talk about these things in the theoretical (the original poster linked to a story that crunched numbers to show it should be safe). My question is does anyone have solid experience they can point to show that it has actually been safe for say 6-18 months under some well known duty cycle (A database, a file server, an e-mail server).

I have actual experience, with crappy flash made by a low end manufacturer that shows me, it's not terrible reliable. It is my understanding that we've had better luck with other makers, but their parts were too expensive (but software development is free *sigh*).

There are other threads in here that make me want to cram a CF-IDE converter into my machines and try putting my journal onto a Flash drive. Sounds like the performance boost and power consumption is a big win, but the fact that every byte of data pushed to the journal might be an issue. On a home machine, it might be worth playing with for giggles for performance testing.

Other folks I know who have tried to do things with flash have also been disappointed over the past 12-24 months, despite assurances from various experts that "it should work"... I'm looking for, "I'd done it, here it is, go play with it.". Now obviously MP3 players have been doing it for a while. I'm more interested in general purpose usage of a Flash drive. Those are the types of things I'm currently working on, cramming a flash into a machine that runs an ext2/ext3/xfs/reiserfs/jfs or some other read/write heavy usage ready FS on it.

Kirby

Re:Longevity of NAND flash

[Anonymous Coward]

I am embedded systems engineer with 10+ years engineering experience with flash devices, fpgas and PALs (and gals) in various projects.
I personally have seen a lot of "failed" flash devices. One of the first Intel PQFP designs I had "ringed with gold" and turned into a necklace because it was so much fun de-soldering SMT components!

Even with the new flash tech, I wouldn't buy a SSD based on flash tech and especially with what is lurking around the corner.

I am *eagerly* awaiting Hitachi's ferromagnetic SPRAM memory tech. It's fast in all respects and does not have the flash write limitations.
Watch and see, this stuff is the death of the IO bound spinning platters folks! Good riddance!
I am just hoping that Hitachi realizes their spinning platters business which has been losing money since day one may be saved by innovation and may be more valuable for it's namesake than it's non-solid state design tech. They need to start building SPRAM based SSD's.
http://www.hitachi.com/New/cnews/070213.html

I purchased a 100MEGABYTE Conner Peripherals IDE harddrive for $850.00 back in the day (I have probably burned more than $200k on junk that I wish I could convert back into cash again). I am tired of being an early adopter and paying exorbitant prices for new toys. These flash SSD's are the new "Conner 100MEG Hard Drive" for me. No thanks...

Re:Longevity of NAND flash

[TooMuchToDo]

Do RAID1 on the flash drives. You may lose data on one drive, but not both. And you'll stick get huge power savings (no spinning disks). As reliability improves, you could do away with RAID.

Re:Longevity of NAND flash

[gallwapa]

my mp3 player had this problem: It was 512mb internal...I used it for a few months (only transferred songs three times in bulk loads) and all of the sudden it was 488mb. grr...

Re:Longevity of NAND flash

[Anonymous Coward]

install all your never changing apps and games to the flash drives for faster loading, and put your swap files on normal HD's. stripe the HD's if your os cant do auto swap striping.

Re:Longevity of NAND flash

[phoenix321]

The capacity problem will be alleviated in 2-3 years, maybe sooner. 900GB archives will remain on spinning disks in the meantime, but their reliability is less-than-ideal. They can fail on you at any time, no matter how well they were treated, how low the operating shocks were and how low the environment temperatures were - they can collapse on the next boot-up, leaving you to employ expensive data rescue operations or simply write off the data and start from scratch.

Re-read the Google report on hard drive failures sometimes, it clearly proves that more than a third of all HDDs fail without prior SMART errors, meaning complete loss of data without a warning. With no backups, you're hosed. SMART keeps track of more than a dozen variables and is usefull in only two thirds of all cases. Thank you, but I'll flock to flash volumes. They need to track only a two variables concerning wear and failure, namely their total power-on hours and their remapped sector count. The first still is just a statistical indication, but the second provides a proportional measure of how much "juice" is left in it.

This is just begging for the usual real-world comparison: bike parts made from carbon-fiber materials. Many advantages, but the definite danger of a sudden and catastrophic breakdown without a prior warning. It may never happen, but if it happens, you'll go down like Brannigans Law: hard and fast.

Re:Longevity of NAND flash

[schon]

When was the last time you actually searched for files you hadn't accessed in six months? Last weekend, when I decided to go through and clean up my home directory.

Re:Longevity of NAND flash

[shicaca]

So what's a ^2 or ^3 between friends?

Re:Longevity of NAND flash

[vaporland]

I had a client, a graphic design studio, that used the same Apple external 120MB SCSI drive from approximately 1992 until around 2002. It was connected to three different file servers over the years, and was powered up for all but probably 60 hours during the entire time. We finally retired it when they switched to a USB iMac for a new server, and it had never been reformatted.

Re:Longevity of NAND flash

[BBandCMKRNL]

Remember on floppies the little plastic that you could move to enable write only? Think about that, but in software. If you hook up an iPod, the in-dash system should automatically play music from it. You should be able to go into a menu and copy from the iPod to the in-dash system, with it asking you to confirm before proceeding. Remember, we're talking teenagers here. "Hey cool, I'll just upload my mp3 player to the car and not even have to plug it in again to hear my music!". A little while later, "Hey Beau, let's upload your mp3 player, too."

Re:Longevity of NAND flash

[TooMuchToDo]

What your talking about is an educational issue, not a technical issue. It's easily fixed in the case you present.

Re:Longevity of NAND flash

[cheater512]

Like I said, if you need very fast speed or high reliability, go for SSD.

However a spun down archive raid array isnt going to suddenly fail like you state it will.
Even with constant use it needs to run for many years before there is a risk of failure and then you'd be prepared for it.
Also if anyone has a large amount of critical data, they'd be using RAID 1, 5, 6 or 10.

SSD wont catch up to spinning disks in terms of capacity any time soon.
Dont hold your breath for a terrabyte SSD hard drive.

Re:Longevity of NAND flash

[peragrin]

you shouldn't let your standard Hard drive become more than 80% full anyways. why? because modern file systems defrag themselves and need the space to separate to be able to defrag themselves.

If your constantly using more than 80% of a Hard drive go buy a bigger one.

Re:Longevity of NAND flash

[Anonymous Coward]

I just looked and you are absolutely correct! The 1997 drives are in the old dell server using Ultra-2 SCSI I have only 10 of those and NO failures( 6 hot in server, 4 sitting as spares). they are all 9.1Gig drives.
You know you forget what is in where when you dont have to touch the stuff for years at a time. Why cant the new servers run forever without problems?

Re:Longevity of NAND flash

[Eddi3]

If you have two identical flash drives, writing identical data, isn't it conceivable that they would fail at similar, if not identical times?

I don't think I'd trust RAID1 in this case. Maybe RAID5? Or maybe a main drive, with another slightly larger drive in RAID1.

Re:Longevity of NAND flash

[phoenix321]

I will hold my breath, really. 120GB is all I need for my main volume, everything else is in network attached storage. That is grid-connected, RAIDed and down in the basement. For a NAS in the basement, I don't really care about individual media reliability, power consumption or noise as long as they are within reasonable eco{nom|log}ical limits. But my notebook, the terminal where I work and have unfinished documents, has to be quiet, light and highly reliable. Backups can get me only so far, as I'll lose some days or hours worth of productivity, a few Euros for a new HDD and maybe a failed presentation at a client. Spinning discs will be on the way out for me the instant a 120GB flash volume of reasonable quality hits below the 500 Euro limit.

Oh and HDDs stored spun-down can fail as well, they'll probably spin up, but hang on clicking at POST. Been there, done that, sorry.

And the storage size argument is not complete: critical data requires a RAIDed setup, thereby "losing" between a third and a half of the installed capacity. If flash based volumes can be as reliable without RAID, they have a 100% advantage versus spinning disks of equal capacity. The economical terabyte of reiable storage as of today requires three 500GB HDDs in a RAID5 setup, meaning 350 Euros for the HDDs, 300 for the enclosure and maybe 100 for the controller. All in all this amounts to 800 Euros and a power consumption I don't like for my SOHO environment. Today, a 64GB SSD volume is 800 Euros as well, so it is 16 times more expensive. Well see how this develops... :)

Re:Longevity of NAND flash

[TooMuchToDo]

The drives won't fail in identical places, nor at exactly the same time (as failure rates are estimates and each drive is slightly different due to manufacturing processes)

Re:Longevity of NAND flash

[plague3106]

No, but they'll fail either reading or writing over time regardless if you are writing or just reading just because the drive is moving. Even if you cool your standard drive, eventually it could just fail because it was left on for 10 years (since an active drive is constantly spinning).

See, that's the thing. I end up replacing the drive long before it fails because it runs out of capacity sooner. I've had one HD die on me, and a few at the workplace. Otherwise though, hundreds and hundreds of these things run for a long time without problem.

I'm all for SSDs, but I'm not sure I want to worry that the capacity is shrinking with each write. I run out of room fast enough with current drives.

Re:Longevity of NAND flash

[Khyber]

Oh, about a century or so in potential lifetime.

Re:Longevity of NAND flash

[cecil_turtle]

I can't believe people are still having this discussion. I haven't seen numbers of writes as low as 100,000 in a few years now. The latest numbers I've seen have been in the low millions, like 2-5 million writes. I wish people would stop citing numbers from 2001 like there's been no improvement in the technology. SSD is coming, people. It's here already for specialty applications, is approaching a reasonable level of price/space/performance for consumer applications, and enterprise use won't be far behind. Disk drives certainly won't be completely replaced in the process as their technology will improve as well, but for applications where pure storage space isn't the problem (performance, power) I think rotating disks will lose out very soon. If you seriously think flash deteriorates appreciably with use then I just have three words for you: click, click, click...

With low enough cost

[jafiwam]

I could do with a 64 GB primary drive on my gaming machine.

Disk performance it the main roadblock to getting on the server first, which has a huge advantage over slower-loading players.

Yes, I am a LPB. Sue* me.

* By "sue" I mean attempt to frag.

Re:With low enough cost

[everphilski]

use a RAM disk for your game?

Re:With low enough cost

[Anonymous Coward]

Excuse me, but where do you live?
Sincerly,
Duke Nukem

Re:With low enough cost

[Thaelon]

Cease and desist having a low ping and entering games first.

Troll, Troll, Troll, and Troll
Attorneys at Law

Re:With low enough cost

[Anonymous Coward]

1st thats a insanely poor joke, second ram cache wins fast hdds any day of the week for almost all games anyways.

Re:With low enough cost

[Anonymous Coward]

The summary said "With the exception of capacity, the solid state disk drives appear to beat spinning disk in every category".

Did they mean to exclude price, too?

Oh, slashdot summary, I trusted you so literally! Why have you let me down?

Reliability

[RonnyJ]

It's not mentioned in the summary, but added reliability might make these types of disks more appealling too:

The no-moving-parts characteristic is, in part, what protects your data longer, since accidentally bumping your laptop won't scramble your stored files. Samsung says the drive can withstand an operating shock of 1,500Gs at .5 miliseconds (versus 300Gs at 2 miliseconds for a traditional hard drive). The drive is heartier in one other important way: Mean time between failure is rated at over 2 million hours, versus under 500,000 hours for the company's other drives.

Re:Reliability

[ElizabethGreene]

They will sell these by the case to the antarctic research station and mountain climbers. They go through normal hard drives like pez because of the cold and low air density. How is the power consumption compared to rotating drives? -ellie

Re:Reliability

[Anonymous Coward]

"How is the power consumption compared to rotating drives? -ellie"

Lower. RTFA, ellie...

Re:Reliability

[ColdWetDog]

They will sell these by the case to the antarctic research station and mountain climbers. They go through normal hard drives like pez because of the cold and low air density. How is the power consumption compared to rotating drives?

I've been climbing mountains for some 30 years. I've never thought to bring a hard drive with me. I've dragged around quite a passel of other odd and heavy things, but I appear to be missing something again ...

Re:Reliability

[Just Some Guy]

Samsung says the drive can withstand an operating shock of 1,500Gs at .5 miliseconds (versus 300Gs at 2 miliseconds for a traditional hard drive).

I was just thinking the other day that 300G just wasn't cutting it anymore. I can't count how many times I've thrown my laptop out of the space shuttle and the drive was barely readable after it landed in a concrete parking lot.

Re:Reliability

[jcnnghm]

The numbers seem pretty reasonable to me. If you drop a drive from a height of 1 meter, it will impact the ground in around ~.45 seconds, with a velocity of about ~4.38m/s. If the drive takes 2 milliseconds to decelerate, the rate would be ~2,190m/s/s, or 223.5Gs. It's not terribly difficult to imagine scenarios where this acceleration could be easily exceeded, for instance by dropping the drive from a slightly higher height or onto a harder surface, like concrete. Even if it does take a full 2 ms to decelerate, there could be a peak of well over 300Gs.

Re:Reliability

[Anonymous Coward]

Hmmm... someone has been a naughty boy and skipped Phys 101. :)

Re:Reliability

[Anonymous Coward]

Yeah, 300G is likely a fall onto a hard surface that will definitely break the outer plastic casing of the laptop, and possibly damage the LCD.

I'm more interested in the life span of 2.5" disks under moderate, but frequent shocks. I've had 3 laptops, and 3 dead drives between them (first laptop none, second laptop had one, and third one had two die). I've owned a lot more desktop drives, and so far only two have failed on me while still in regular use. A laptop drive takes a lot more abuse, and I'd be curious to know how much daily shocks in the 5-10G range affect the failure rate. Presumably, such small accelerations would have no impact at all on a flash drive.

Re:Reliability

[Hucko]

phhhtttt. Amateur.

Re:Reliability

[Debug0x2a]

With the crystal sapphire LCD that ASUS has put out, i think a combination of that and solid state drives could make for some extremely durable computers.

Re:Reliability

[ElizabethGreene]

Curious. You don't take your laptop everywhere? I was specifically referring to this story Geek on Everest, which made /. a while back. Also the Antartic data center story from the register this month. :) -ellie

Re:Reliability

[babbage]

I met a well known mountain climber recently who mentioned that he has been bringing laptops to Mt Everest for years now, but they all stop working above a certain altitude due to the air pressure drop (and, to a lesser extent, the temperature). Apparently it has been a big problem for the work they do for years now.

At altitude, the hard drives basically seize up, and the LCD displays develop a faint spiderweb pattern that makes the picture difficult to discern. Once they come back down, things start working normally again -- the hard drive starts spinning again and the LCD goes back to normal. Above a certain level though, the traditional laptop technology just doesn't work right.

He mentioned that they used to use original hard drive based iPods for data collection they do at the summits, but the flash based Nanos are much more reliable for their needs (not to mention smaller, lighter, and longer battery life...). Once the laptops start shipping with some kind of reliable solid state storage, they'll migrate to those also.

and ...

[Billly Gates]

... about 2 megs per second under Vista.

WIth 1 million writes per second before the ram goes bad I would be worried about reliability. I hope the firmwire at least maps out bad sectors.

Hmm

[orclevegam]

I'm really interested in the SSD drives as high performance replacements (particularly for holding OS images where boot times should be nicely reduced), but I've got to wonder how the mean time to failure of one of these compares to a traditional magnetic disk. I know they use write leveling, but that just means everything will have a tendency to fail around the some time later, rather than a spot or two now and then. Anyone have any actual reports on these? I can usually make it 2 or 3 years before I start to see errors crop up on magnetic disks (sometimes more or less depending on how much thrashing the disk is subjected to). Might it be cheaper to simply buy a decent sized CF or SD card and an ide/sata adapter rather then paying for an actual disk, or is there some inherit advantage to one of these you'd be missing out on?

Re:Hmm

[orclevegam]

Before someone points it out, yes I know the article quotes some mean time till failure for the drive, but I'm wondering if anyone has any actual experience with these, not what the marketing department says the performance should be.

Re:Hmm

[jandrese]

No, no reasonable people can afford them yet.

Re:Hmm

[Bandman]

I don't think they've been around for the timespan they say they're good for yet, have they?

Re:Hmm

[Sibko]

If you had RTFA, you'd probably have noticed it said this:

"Samsung says the drive can withstand an operating shock of 1,500Gs at .5 miliseconds (versus 300Gs at 2 miliseconds for a traditional hard drive). The drive is heartier in one other important way: Mean time between failure is rated at over 2 million hours"

Re:Hmm

[orclevegam]

Please see this post.

Re:Hmm

[Toveling]

CompactFlash cards aren't going to be anywhere near as fast as this, even the high-quality cards. Most top out at 20mb/s.

Re:Hmm

[DamonHD]

I already boot my low-power Linux server off CF:

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

Rgds

Damon

Re:Hmm

[orclevegam]

Most top out at 20mb/s. Well, if the metrics quoted in the article are to be believed the actual performance on the drive they tested averaged 25mb/s with burst speeds of 30mb/s rather than the manufacturer quoted speed of 100mb/s, so if you can actually get 20mb/s out of a CF card, that's not much of a performance hit. Happen to know what the actual mean and burst read speeds are on a traditional HD (more concerned with reading then I am with writing, would be using the CF drive mostly for app/OS storage with media and data on a traditional HD)?

Re:Hmm

[DoofusOfDeath]

I know they use write leveling, but that just means everything will have a tendency to fail around the some time later, rather than a spot or two now and then.

Not necessarily. It really depends on the statistical distribution of the number-of-writes-until-failure the various blocks (or whatever the unit of failure is) in a SSD. If they're normally distributed, then you'd probably see several blocks fail here or there long before the majority of them had failed.

OTOH, if you or your operating system are never told that blocks have begun to fail here and there, then you might not be aware of the device's poor health until lots of blocks have failed. This might be exacerbated if a SSD silently handles failed blocks under the covers.

Re:Hmm

[archen]

Good luck using that CF card without DMA though. Heck I've installed OSs that simply barf trying to use CF as a hard drive because they can't figure out what to do without DMA. In my experience CF cards are good for dedicated firewalls and such that don't do much disk access. They're also not bad for storage as they're rather durable compared to the alternatives. Aside from that I wouldn't really use them for much of anything. And I doubt most CF cards bother with write leveling unless you get the industrial ones.

SSD drives on the other hand do write leveling, and support DMA. I haven't really seen them pull ahead much in my own testing as far as performance, but the low power and durability are mainly what I use them for.

Re:Hmm

[Anonymous Coward]

There are many converters (IDE->CF) that support DMA. I've used several and they get about 20mb/s with the beest CF cards. Prestico Card Master IDE-811CF/R is the best and cheapest i've found with DMA support. (They also have a SATA version if you need that)

Re:Hmm

[jdjbuffalo]

I don't know all the details of these drives but I would think that they would be able to produce some S.M.A.R.T. errors that could tell a diagnostics program of what shape it's in in terms of failed sectors (maybe needs a new revision for this feature).

Re:Hmm

[Anonymous Coward]

This issue has been hashed out here on slashdot more than a couple of times. With write leveling the MTBF will be higher on an SSD than on a traditional drive. Moreover, those failures will be far less costly because only writes will fail, not reads. So when the drive fails, all you have to do is buy another one and copy everything off the failed drive.

At this point, there are 2 reasons and only 2 reasons left to not choose an SSD...price and capacity. Granted, these are pretty important considerations, but if flash can narrow or eliminate the gap in those two areas, there won't be much of a reason to go with traditional hard drives. Even with those considerations, flash still makes a lot of sense for situations where durability and power consumption are important.

Re:Hmm

[IrquiM]

Erh, CF cards are not that expensive ;)

My sattelite tuner uses one as /

Is it just me?

[crymeph0]

Or does the linked article say nothing about TB sized drives, only the flash drive?

Re:Is it just me?

[cthulu_mt]

You are correct. Way to fail /.

Re:Is it just me?

[bcrowell]

No, it's not just you. The /. summary seems to bear little resemblance to the actual article. There's also no mention of the pricing or availability of the SSD, but from a quick check on frys.com, it looks like it's not available yet, what is available is 32 Gb sizes, and 32 Gb sizes will set you back about $350.

Re:Is it just me?

[Ruddykins]

Naw, I wondered about that too, but it wasn't worth it to me to post about it. :)

Re:Is it just me?

[Jesus_666]

With a little digging around in the "Hardware" section you can find three Terabyte HDD reviews, one for "The 1TB Barracuda", one for the WD RE2-GP and one for the Hitachi Deskstar 7K1000.

Interestingly, the Seagate has so much space that "[t]he odds are excellent that Windows will never again tell you that you're running low on hard disk space with this 1TB drive, and that alone might be worth the price of admission", while the equally-sized Hitachi "doesn't boast efficiency, but its slightly lower platter density allows it to achieve better error-checking without the need for sophisticated firmware". Either Hitachi's drive is somehow bigger than Seagate's or Hitachis just fill up faster.

Re:Is it just me?

[xZgf6xHx2uhoAj9D]

Here you are, your majesty.

Re:Is it just me?

[Anonymous Coward]

[t]he odds are excellent that Windows will never again tell you that you're running low on hard disk space with this 1TB drive

As opposed to all those other 1TB drives that you're likely to fill up in no time.

Re:Is it just me?

[oKtosiTe]

Yes, it is.

Re:Is it just me?

[Anonymous Coward]

http://www.computerworld.com/action/article.do?command=viewArticleBasic&taxonomyName=storage&articleId=9052381&taxonomyId=19&intsrc=kc_feat

scroll down...

Re:Is it just me?

[cnettel]

Seagate gets to 1 TB with a lower number of physical platter( side)s. They all fit in 3,5-inch enclosures, but that still doesn't make them equal. However, equating reliability with (lower) platter density is quite strange, especially when we get into the matter of one manufacturer using perpendicular recording.

nice review

[theoverlay]

There are a couple of great articles about the benefits and drawbacks of ssd at http://www.infiniteadmin.com/

Number of writes?

[QuietLagoon]

With the exception of capacity, the solid state disk drives appear to beat spinning disk in every category,

Why is the ultimate number of writes never taken into account in these comparison reviews? Why are solid state drives tested so that their weaknesses are not probed?

Re:Number of writes?

[Planesdragon]

Why is the ultimate number of writes never taken into account in these comparison reviews? Why are solid state drives tested so that their weaknesses are not probed? Because it's a measure best reflected by Baysean Data, and they don't have enough time to test them.

If you want, buy an HDD and a Flash-Drive of the same cost, hook them up to a program that runs each at equal data-transfer rates, and see how much data you can read and write to each before they fail. Report back to us in the six months it'll take you.

Oh, and you need to do the trial over a wide sample, so get, oh, at least ten of each.

Re:Number of writes?

[iminplaya]

Because most of these "reviews" are actually advertisements, or at best press releases. The reviewers aren't going to bite the hand that feeds. You'll frequently get better info from user comments, which, unfortunately is mixed in with the dreck.

MTBF/Write Cycles

[Lookin4Trouble]

Since I've seen this plenty of times, I'll address it.

Write Cycles: Even at the lowest estimate, 100,000 write cycles to failure

Meaning on a 32GB Drive, before you start seeing failures, you would have to (thanks to wear-leveling) write 32*100,000 GB, or 3.2Petabytes

at 60MB/sec write speed of the Samsung drives, you would need to write (and never, ever read) for 3,200,000,000/60, or ~53Million seconds straight.

53Million divided by 86,400 means you would need to be writing (and never ever reading) for ~617 Days straight (That's roughly 20 months of just writing, no reading, no downtime, etc...

So... the sky is not falling, these drives are slated to last longer than I've ever gotten a traditional drive to last in my laptop(s)

Almost forgot to mention, standard NAND of late has been more in the 500k-1M write cycle between failures range. 100k was earlier technology, so multiply numbers accordingly.

Re:MTBF/Write Cycles

[jafiwam]

What happens when you run the same napkin math on a drive that has Windows, Office, and two big games on it?

That leaves you about 10 GB of space to use for writes for swap, temp files, etc.

Re:MTBF/Write Cycles

[everphilski]

Meaning on a 32GB Drive, before you start seeing failures, you would have to (thanks to wear-leveling) write 32*100,000 GB, or 3.2Petabytes

NOT true, unless the drive is completely empty! If you have 31 gigs of data on that drive which you were using as long-term storage, then you'd only have to write (32-31)*100,000 GB of data before failure. You obviously wouldn't be overwriting any data already stored on the drive ...

Re:MTBF/Write Cycles

[goofy183]

Except most wear-leveling MOVES data around on the drive. Since random access is 'free' shuffling mainly read-only data around on the disk periodically is perfectly reasonable.

Re:MTBF/Write Cycles

[NormalVisual]

There's a serious flaw in your analysis - you're assuming a totally empty drive. You're going to be wearing the drive more and more as it gets full, and the combination of an almost-full drive and a busy swap partition might get interesting very quickly.

I agree that on the whole, flash is a lot more durable now than it used to be, but I'm not quite convinced that these will be suitable as a general-purpose replacement for magnetic disks. Aside from the NAND longevity issue, I'd be concerned about the ability to recover data in case of a controller failure or other hardware-related issue. Mag disk is relatively easy to deal with in that regard.

Re:MTBF/Write Cycles

[baywulf]

Not true if the drive uses static wear leveling algorithms. These algorithms will swap data between low use and high use NAND regions periodically.

Re:MTBF/Write Cycles

[Anonymous Coward]

This is obviously true only if your mind can't conceptualize a swap operation.

Re:MTBF/Write Cycles

[vadim_t]

It's still not the same failure mode though.

On a magnetic hard disk, once you get a failure you can expect the thing to die completely soon, because failures tend to be mechanical. Once there's scraped magnetic material bouncing around on the inside it's only going to get worse, possibly very fast.

On a SSD what should happen is that sectors die in a predictable fasion, and they die due to writes, so you can still read and recover your data.

Re:MTBF/Write Cycles

[NMerriam]

You obviously wouldn't be overwriting any data already stored on the drive

No, but the wear-leveling routines in the drive will happily move around your existing data so that rarely written sectors are available for heavy writing operations.

Seriously, this "issue" comes up in every discussion about SSDs, and it seems like people are just unwilling or unable to accept that what was once a huge problem with the technology is now not even remotely an issue. Any SSD you buy today should outlive a spinning disk, regardless of the operating conditions or use pattern. It is no longer 1989, engineers have solved these problems.

Re:MTBF/Write Cycles

[renoX]

Please mod parent up! I'm sick of all these posts (modded up!!) who thinks that writing on a mostly full disk remove the effectiveness of wear-leveling, there is no reason why this should be the case..

Re:MTBF/Write Cycles

[Anonymous Coward]

That's exactly what I was thinking. Sadly, the wear leveling mechanisms aren't documented in any meaningful way whatsoever. The most important question is, how the drive can determine which parts of the disk are "empty". If it cannot, then, after filling the drive just once to its maximum capacity, The number of write cycles left for will be only (combined size of hidden spare/wear leveling sectors)*100,000. Will that also happen if you use an "unsupported" filesystem, and, if there is such a thing, which filesystems are supported?

Re:MTBF/Write Cycles

[Zerth]

Smart wear leveling enables the drive to swap files on sectors with many writes left(i.e., read only or rarely changed files) and with those on sectors with few writes left(swap, savegames, etc).

So performance isn't that far from a nearly empty drive.

Although I do agree, I'd be concerned about recovering from controller failure more than with a magnetic drive.

Re:MTBF/Write Cycles

[jafiwam]

No, but the wear-leveling routines in the drive will happily move around your existing data so that rarely written sectors are available for heavy writing operations.

Seriously, this "issue" comes up in every discussion about SSDs, and it seems like people are just unwilling or unable to accept that what was once a huge problem with the technology is now not even remotely an issue. Any SSD you buy today should outlive a spinning disk, regardless of the operating conditions or use pattern. It is no longer 1989, engineers have solved these problems. Actually, I think the issue is there are differences in the drives that don't come up in the articles themselves, so that detail gets left out every time.

So, it's inevitable that someone who doesn't know this particular detail, but is already familiar with how platter based magnetic media work will come up with that issue in pretty much every discussion.

The problem is it's new. That's all. (Or, perhaps that techno-journalists write about stuff they don't know enough about.)

Re:MTBF/Write Cycles

[pushing-robot]

Why would anyone use flash for virtual memory? You can get 4GB of DDR2 SDRAM for seventy bucks, or two gigs for less than half that. Notebook SO-DIMM prices are about the same.

With DDR2 prices so cheap, I don't see why anyone (with a modern enough system to use DDR2) is swapping data to disk regularly. Certainly not anyone who can afford a SSD.

Re:MTBF/Write Cycles

[orclevegam]

Well, it may not entirely negate the effectiveness of wear leveling, but it definitely makes the calculations a bit more complicated. Lets look at the theoretical example of a 32G disk with 31G used and a 512M write about to happen. It decides that the free space already has too many writes, and it needs to write the data to a used section of the disk instead, so it finds a 512M chunk of data that has the lowest write count and copies that to the free space (with a high write count, further increasing it's write count). It then takes the new chunk of data and overwrites the old chunk of data (once more increasing a 512M blocks write count). Now, there's only two issues here. First, even though you've moved some data which you hope is going to be semi-permanent (as it had low write counts) to a high count section, there's no guarantee that the user won't turn around and delete those files in the next couple minutes making it pointless to have copied the data there. Second, to write 512M of data, you've just had to perform 2 512M writes thus using twice the cells to store the actual content. Yes, following this strategy should increase the time till you see a failure, but it also invalidates the simplistic calculation used to come up with the 50+ years till failure figures some people have proposed. Now, I'm not saying the performance won't still be a long time, just that it's not as cut and dry as it might seem.

Re:MTBF/Write Cycles

[TooMuchToDo]

As long as the driver is smart enough to disable a paging file, not that much writing is done to the hard drive on a Windows box (at least by the OS). When you do updates of course, writing is done. And when you save files, writing is done. But if you're just surfing the web and have 2-4GB of ram, disable caching, and the browser shouldn't write to the disk. If you're running Office, or games, save your work or savefile over webdav to a remote provider or use Amazon S3 to save those small amounts of data.

Re:MTBF/Write Cycles

[Skapare]

If the block on disk has ever been written the flash device has to keep it. It has no idea that no file inodes point to it anymore. Wehn a write is done, it picks a block from the pool and writes it there, and juggles its own mapping. But I am curious about a flash device that will, on its own, just juggle things around. That could avoid the data stagnation problem, where any data that doesn't get written on is just keeping the zones of writing all that much smaller. But it can also increase the number of writes being done. It would have to know that certain blocks don't get frequently written for that to be effective.

Re:MTBF/Write Cycles

[orclevegam]

It's independent of filesystems as the wear leveling happens at a much lower level. It keeps track of writes per cell, which are physical locations, not logical (not to be confused with physical and logical as is used when programming, this is lower then that) and maintains logical mappings at the PATA/SATA interface. It doesn't know anything about the data, just that the data is being written, so it finds a chunk with low write count and writes the data marking the physical location as being mapped to the logical one requested. What interests me of course is how the physical/logical mapping is maintained, as obviously some sort of storage must be used which itself should have a MTBF. Who knows, maybe the logical mapping is stored in the cell itself, which would be interesting but I think would also create a "seek time" as it would require a scan (or cache) of the cells to find a particular logical address.

Re:MTBF/Write Cycles

[Anonymous Coward]

> What interests me of course is how the physical/logical mapping is maintained, as obviously some sort of storage must be used which itself should have a MTBF. Who knows, maybe the logical mapping is stored in the cell itself, which would be interesting but I think would also create a "seek time" as it would require a scan (or cache) of the cells to find a particular logical address.

In one wear-level scheme a swap/remap might occur only if a highly-used block is 1000 writes more worn than the least worn replacement-to-be block. In such a case the physical to logical map would be written-to much less frequently than the regular storage, thus having a longer MTBF.

Re:MTBF/Write Cycles

[Just Some Guy]

Even at the lowest estimate, 100,000 write cycles to failure

Hey, I never get this question answered: the bad block map has to be stored somewhere, so is it also limited to 100,000 writes? You can't remap the map, can you? If not, then, are you limited to 100,000 total errors?

Re:MTBF/Write Cycles

[settrans]

Even assuming that you are correct about the longevity of these devices, you're still assuming that the drive is empty when you start writing, and when the drive is full, you continue by overwriting the drive sequentially ad infinitum. If the drive began 99% full to begin with, and you're recording and rotating logs, say, for a web server, the situation is very different...you have the equivalent of a drive 1% of the capacity to wear level over.

Re:MTBF/Write Cycles

[larry bagina]

Meaning on a 32GB Drive, before you start seeing failures, you would have to (thanks to wear-leveling) write 32*100,000 GB, or 3.2Petabytes

Data isn't written by the byte, it's written by the block (512 -- 4096 bytes). And don't forget that wear leveling requires swapping blocks (and therefore 2 write operations instead of one). And don't forget that writing 1 byte (potentially) causes just as much wear as writing an entire block.

Re:MTBF/Write Cycles

[sxeraverx]

That would be true if you didn't take into account a full, or nearly-full disk. To be able to test this thing, at a ridiculously high estimate, say, 10M writes, the napkin math would be as follows: 32GB initial write 10M writes x 1024kB/write = 41.5GB At even 1MB/sec, that's 42,500 seconds or about 11.8 Hours straight. That's roughly half a day of just writing, no reading, no downtime, etc. At 60MB/sec, that's 12 minutes. Compare that to Seagate ES2's 1.2 Million hours MTBF (for platters constantly spinning), and you'll see, the sky really is falling.

Re:MTBF/Write Cycles

[Anonymous Coward]

Is this your assumption or do you know for a fact they do this?

On a typical storage system you'd fill to a certain level with your applications and then write and delete the last portion of the free space.

So these drives are constantly moving around the data already written? Otherwise the 100,000 write in the last 100MB of free space will take a significantly shorter time to degrade.

To those worried about longevity...

[ichigo 2.0]

((64 GB) / (80 MB)) * seconds * 1 000 000 = 25.959436 years

What about real performance

[B5_geek]

How do these SSD compare to a real high-end disk like a 15k rpm Ultra320 SCSI drive?
Of course SSD will beat an IDE disk hands-down, but that is not why you buy IDE drives.
I have always used SCSI for my OS/system and IDE for my storage, this combination (in addition to SMP rigs when available) has allowed me to out-live 3 generations of processors. Therefore saving me money on upgrades.

SSD seems best marketed to 'gamers' so why is it always connected to a very limited IO bus?

Re:What about real performance

[jd]

I've seen SSDs used to cache access to a traditional hard drive, and have even seen traditional hard drives used to cache access to optical mass storage. So long as your disk usage is "typical" (lots of access to a limited range of programs and libraries, infrequent access to the full range), it makes sense to layer the access in this way. You don't then have to care about limited space on the SSDs, you don't have to worry about MTBF because it's very unlikely all layers will fail at the same time (cacheing means that you've spare copies of everything that you frequently use), and you're much less subject to the limitations of any of the devices.

As for the bus, I would have to agree that most Flash-based or other popular SSD solutions are far slower than they need to be. (Bear in mind that PCI Express 2.0 can deliver 5 gigabits per second per lane, across 32 lanes. Yes, not many home PCs have PCI Express 2.0, but even conventional PCI technology is capable of delivering more than adequate bandwidth. To make the SSD faster, it could also be cached using conventional RAM. Ideally, it wouldn't be a write-through cache, but a battery-backed standard cache that flushes as-needed or on external power failure, so as to minimize writes.

As with most technologies, if there aren't good solutions it is not the fault of the technology, it is the lack of imagination by those implementing it.

Re:What about real performance

[KonoWatakushi]

No need to compare with 15k rpm drives; flash disks lose spectacularly to low rpm laptop drives for random write performance. For obvious reasons though, no one ever tests random write performance. Manufacturers also rarely report random write IOPS.

Flash is great, if your disk is basically read-only.

Re:What about real performance

[Bearhouse]

Indeed - good for booting your OS from, then, as another poster has pointed out.

Re:What about real performance

[DDumitru]

Random write performance to bare drives is usually quite bad. Most "reputable" vendors do publish random write figures. SanDisk quotes 13 IOPS in the spec sheets. Mtron quotes 120 IOPS. I have not seen quotes from Samsung, but have tested their old drives at 27 IOPS. I even tested one drive at 3.3 write IOPS.

On the other hand, random writes issues are "fixable". My company just published tests for various Raid-5 Flash SSDs setups. For 4 drives testing with 10 threads on Linux 2.6.22 using our MFT "driver", we get:

4K random reads 39,689 IOPS 155 MB/sec
4K random writes 29,618 IOPS 115 MB/sec

These are real numbers and the application does see the performance improvement.

For full details on drive performance see:

    http://managedflash.com/news/papers/index.htm

Re:What about real performance

[Anonymous Coward]

http://www.fusionio.com/ - assuming this ever materializes as a product, it may be something to look at.

Re:What about real performance

[skaaman]

Here is a good quick rundown against a WD Raptor. Also note that testing reveals an issue with the Intel ICH9 and ICH8 chipsets that seems to cap sustained transfer rates at about 80MB/sec. http://www.anandtech.com/storage/showdoc.aspx?i=3167&p=1

Not the jump I was hoping for

[Badmovies]

The new solid state drives did beat out older drives in terms of performance, but I can honestly say that I was hoping for a bigger difference between the two in terms of performance. Not just "beating" the older technology, but beating it by an order of magnitude.

Looking at it, the biggest benefit I can see is that the solid state drives should be better at withstanding shock and vibration - which normal hard drives hate. If they cannot improve the performance (which will still be useful for gamers, servers, and other speed freak things) then reliability and security of data is the selling point. I can see rugged notebooks using these.

Re:Not the jump I was hoping for

[Ant P.]

Nope, the biggest benefit is that they don't have a 5400rpm motor spinning up and down every time you want to get data out. That and the screen are what drain laptop batteries, the CPU these days is relatively efficient.

Re:Not the jump I was hoping for

[TimothyJones]

That's because those are not really performance SSD drives. Random Access time is much improved but the transfer rate is way below a good HD. MTron has some high performance drives that pulverize everything else but they do cost an arm, leg and probably one of your kidneys. The only real benefits of those Samsung SSD's are much lower power consumption, no heat or noise. On a laptop this is still very good news.

Re:Not the jump I was hoping for

[baywulf]

It would not be too hard to increase the sequential performance by striping data across more NAND. Random read performance is also not too hard. The hard part is always random write performance. This is because if you want to modify a sector of data, all the remaining data must be moved to a new place. The copying of old data takes lots of time but tricks can be used to optimize them out but for true random writes, the performance will never be that good with the current NAND limitations.

Re:Not the jump I was hoping for

[Badmovies]

That's a very good point and definitely a factor for laptops. Still, one of the "golden bullets" I want now is a blazing fast way to retrieve data from the storage.

Well, that and size. Give me the power of a laptop in something the size of a cell phone, with a projected screen in midair, with a way of registering me typing commands (without a keyboard) and I will be happy with my computer - for a year or two.

Re:Not the jump I was hoping for

[Skapare]

They should be able to parallel several flash chips to increase the speed. Or maybe the old drives already did this?

Re:Not the jump I was hoping for

[MrSteveSD]

I suppose with flash drives there is more potential for spreading things out in parallel and maybe they are not taking advantage of that yet. For example, when copying a 1 gig file, a flash drive could break the file up and simultaneously read from/write to ten different flash chips.

It gets better

[WindBourne]

My home server has a terabyte of disk, but I added a CF-IDE adaptor card, along with 4G CF card. I loaded Linux kernel on it, and then mapped a few dirs to partitions on the HD. After about 6 months at it, I noticed that the temp in the case dropped. It appears to be about 5-10C lower (depending on load). The disk spend the bulk of their time sleeping. I have been pleased enough with this server, that I am going to do the same to my small shoe box computer. Rip out the HD, add CF for /, and then mount my home dir from the server.

Re:It gets better

[Anonymous Coward]

Welcome to winter.

Re:It gets better

[orclevegam]

That's almost the exact same setup I had been considering. If I can scrape together the money I was also looking at maybe getting a small form factor box to setup next to one of my TVs. Going to put a CF card for the OS/Apps, with either an internal or external large HD for media storage and load MythTV on it. Could then use the desktop as a client to the TV box and watch TV at my desk.

Re:It gets better

[i.of.the.storm]

Yeah, as the AC said 6 months is long enough for a complete seasonal flip (perhaps from summer to winter, but you didn't provide any dates so we can't tell) and at least for me, temps drop around 10C from summer to winter so it may have just been environmental factors rather than the CF card. Then again, I do live in California, so perhaps other states don't have as much seasonal differences.

what about number of reads

[josepha48]

ssd's have a limited lifetime of number of reads and writes. I would imagine that a 10 year old hard drive that was used all day would last longer than an ssd. I would also imagine that someone who does a lot of compiling and disk writes would wear down the ssd and then have to throw it out and replace it. I know that they have some technology that spreads this out on some devices, but still. I think having the main OS on an SSD would be ideal and then the swappable parts could be on a regular disk. You could make the OS read only, so it would be less likely to have a hacker install virus software to the OS directly and the OS could refuse to run software in the read only space. While this would be limiting, it would solve some of the issues we have today with virus and security.

Re:what about number of reads

[stewbacca]

I would imagine that a 10 year old hard drive that was used all day would last longer than an ssd.

Considering in over 25 years of computing, I've never seen a hard drive last longer than 5 or so years, I'm just going to go ahead and bet that an ssd would be a better choice for me.

Re:what about number of reads

[Daimanta]

I'm not even 25 years and I have seen PLENTY of spinning hdds grow old. I currently have a 8 year old one running right next to me. Well, I consider myself lucky since I never had the problem of a failing hdd. And by never I mean NEVER.

Re:what about number of reads

[stewbacca]

Consider yourself lucky. There is plenty (albeit anecdotal) evidence just here on slashdot to the contrary. MOST of us have seem to be blessed with the "dies in less than four years" models. Hard drives fail so frequently that I can't even think of how old the oldest one I've ever owned lasted.

Re:what about number of reads

[josepha48]

the hard drive in my toshiba laptop is a 4 gig drive that has lasted 12 years. I have several drives around here that are 2 and 4 gigs that are about 10 years old. My ssd's on the other hand, no so long. Only so many writes you can do to a ssd before it goes to crap, and if you do a lot of compiling you are going to loose your data quicker on an ssd than a hd.

Where can I buy one?

[Sibko]

So... yeah. I want one, I'm sure there's more than a few other slashdotters out there who also want one. But none of Samsung's links helped me find a store that sells the 2.5 inch 64GB drives. Does anyone know where these are being sold?

Re:Where can I buy one?

[baywulf]

Try newegg.com

Re:Where can I buy one?

[Alt_Cognito]

For Samsung, it's easy enough, but the rest of the SSD's look geared toward that enterprise market.

Re:Where can I buy one?

[ricky-road-flats]

Near me, this place has a handful of different ones.

Re:Where can I buy one?

[Anonymous Coward]

Hahahahahah... nice "consumer-friendly" prices.

No one's going to buy these fucking things until they drop to an affordable amount -- affordable means: prices competing with existing non-solid-state drives.

Until then, it's a pipe dream. Even Juniper doesn't use this kind of media in their M20s and M40s -- they use ATA disks (and should be ashamed for it).

Re:Where can I buy one?

[Fweeky]

They'd be reasonable in a server. We just spent £6000 on storage for a single system where we're more interested in IOPS than capacity, and if we could spend a similar amount for much faster storage you can be sure we'd seriously consider it.

It's just a shame most manufacturers seem to be concentrating on mobile users, not those who need serious IO.

real review here

[Anonymous Coward]

This review here compares several Mtron SSDs to the Samsung and a Sandisk:

http://www.tabletpcreview.com/default.asp?newsID=1037

[Spoiler]: They beat the crap out of the Samsung SSD, are available now and already several hundred bucks below a Grand.

Waiting for low-end drives

[alegrepublic]

I am still waiting for a reasonably priced low-end drive. An 8GB usb drive can be found for about $50. Packing 4 of them and replacing the usb circuitry with SATA would make for a 32MB for $200. Granted, it may not be the fastest drive around, but sometimes speed is not the most important factor. A 32MB would be enough for installing any current OS and still have some room for personal files to carry along on a trip. So, I think the current trend of providing high-end drives only is just an attempt to milk users to the maximum without much concern for what we actually need.

Re:Waiting for low-end drives

[skolima]

A 32MB would be enough for installing any current OS and still have some room for personal files to carry along on a trip. We're still in 2007, riiight?

Re:Waiting for low-end drives

[jherrick]

32MB would be enough for installing any current OS Okay, Bill Gates.

Re:Waiting for low-end drives

[Pop69]

If you packed 4 8GB sticks together and ended up with 32MB at the end, what did you do with the rest of the space ?

The REAL article link

[hack slash]

Review: Six top hard disk drives for speed and capacity

Speed for a mech. HD is burst, not track-to-track?

[Futurepower(R)]

Quote from the Computerworld article and the Slashdot summary:

"Samsung rates the drive with a read speed of 100MB/sec and write speed of 80 MB/sec, compared to 59MB/sec and 60MB/sec (respectively) for a traditional 2.5" hard drive."

The speed quoted for a mechanical hard drive is a burst speed, accurate for reading only one track, and doesn't include the time it takes for a conventional rotating hard drive to change tracks. Isn't that correct?

Re:Speed for a mech. HD is burst, not track-to-tra

[0123456]

"The speed quoted for a mechanical hard drive is a burst speed, accurate for reading only one track, and doesn't include the time it takes for a conventional rotating hard drive to change tracks. Isn't that correct?"

Depends. My IDE drives seem to sustain 60-ish MB/second on a large contiguous file even across multiple tracks... but suck if the file is heavily fragmented.

Swap partition/file

[Jamu]

Is it worth getting a small one and using it for swap? In other words: Is it faster than a normal HDD? And how long would it last (with this usage)?

Re:Swap partition/file

[fm6]

I'd guess it'd he faster — but not as fast as increasing your RAM so you didn't swap as much.

Re:Swap partition/file

[johannesg]

Why not just buy enough RAM? It is cheaper than using a solid-state disk, and if all you use it for is swap anyway it really doesn't matter if it volatile or not...

Re:Swap partition/file

[phoenix321]

Some people reboot or power down their systems once in a while. To save power, to silence their bedroom or because of company policy. Every boot-up is several thousand random disk accesses and that's going to take a while in a mechanical disk with several milliseconds access time. With flash-based media of even low to mid transfer speeds, the boot process is a whole lot faster thanks to read access in the nanoseconds.

And most consumer PCs are running windows versions that do not support more than 3GB of RAM. And the next half-life and whatnot will use all those and more...

Re:Swap partition/file

[johannesg]

The grand parent was talking about putting his swap partition on the solid state device. Swap partitions do not get saved across reboots. So thanks for stating the bleedingly obvious, but my question remains: what is the point of having a solid state device for swap space if you can simply buy enough RAM in the first place?

Re:Swap partition/file

[repvik]

Because most operating systems page out unused data even if you have plenty of free RAM. That way the free RAM can be used for cache or a sudden peak in memory usage.

Re:Swap partition/file

[phoenix321]

That and because a certain popular desktop operating system doesn't support more than 3GB of RAM *and* requires regular reboots.

Here ius a 461GB one ...

[foobsr]

... not affordable, of course.

But that is what you get in the not so distant future:

* Performance:
o Access time -- 30 to 100 microseconds
o Burst transfer rate -- 300 MB/sec.
o Sustained transfer rate -- up to 100 MB/sec.
o I/O operations per second -- Up to 20,000
* Environmental specifications:
o Operating temperature 0-70 degrees C ("commercial" range), -40 to +85 degrees C ("industrial" range)
o Shock (operating) -- 1,250 G
o Vibration (operating) -- 16.4 G rms
* Reliability:
o MTBF -- 1.9 million hours, minimum
o Error correction -- corrects up to 9 random bit errors per 528-byte block
o Data integrity -- up to ten years
o Read endurance -- unlimited
* Physical specifications:
o Form-factor -- 2.5-inch HDD
o Dimensions -- 2.75 x 3.95 x 0.93 inches (69.85 x 100.45 x 23.55 mm) maximum, storage capacities 64 GB and below are 0.33 inches thick
o Weight -- 2.9 to 7.8 ounces (83 to 221 grams)

CC.

Then don't fill the drive

[tepples]

This is always claimed as the solution, "evening" writes. But I think the question about how long will the drive last is still relevent; all it takes is a mostly full disk, which has a high I/O load. Easy: don't let the drive become mostly full. This means heavy-duty drives will be a 64 GB chip reformatted for 48 GB with the rest designated as spare sectors for wear leveling, but the power consumption and seeking speed benefits can still make it worthwhile.

Re:Then don't fill the drive

[plague3106]

Easy: don't let the drive become mostly full.

Ya, becuase THAT is realistic in the real world...

Re:Then don't fill the drive

[DRAGONWEEZEL]

While not the most important, it can have a dramatic effect on level loading times.

Re:Then don't fill the drive

[DRAGONWEEZEL]

You shouldn't let a drive become mostly full. When a hard drive has 30% (I think it's around there...) space it starts to become less efficient.

Ideally, you could just lock down the read only portions to flash, and let the harddrive take care of the write intensive work.

Re:Then don't fill the drive

[cecil_turtle]

It is, actually. Any competent server admin won't let drive space drop below ~30% for performance / fragmentation reasons. For a new system, accounting for growth + free space overhead, I normally shoot for something like 70-80% free space. E.g., if I have 100GB of data to store now, I will allocate 400GB of storage for it and then start looking to add more space or split the data up when it hits the 280-300GB in use ballpark.

Bayesian or Monte Carlo?

[mosel-saar-ruwer]

Because it's a measure best reflected by Baysean Data, and they don't have enough time to test them.

What's Bayesian Data? [And yes, I am too lazy to Google it.]

Did you mean Monte Carlo?

Or maybe Latin Squares?

Re:Speed for a mech. HD is burst, not track-to-tra

[Anonymous Coward]

No, burst would be reading from the cache in the drive, which is typically done at near interface speed (several hundred megabytes/s).

With the Exception of What???

[Nom du Keyboard]

With the exception of capacity, the solid state disk drives appear to beat spinning disk in every category,

Well excuse me, BUT, capacity is the single largest factor in my disc drive purchase decisions. I'll give away speed, power consumption, size, heat, noise, and even cost - everything but reliability - in favor of capacity. Even "slow" hard drives are quite fast historically speaking, and none of those other factors make up for running out of drive space.

And don't the SSD's cost a lot more too? Capacity and Cost, the two biggest factors to consider.

Re:With the Exception of What???

[Courageous]

It's not all one way or the other. At some price point, there are consumers who will buy the fast thing, just to have SOME of the fast thing. One is not required to have only one thing inside one's computer. For example, I could make productive use of a 32 or 64GB flash system if it were fast enough and cheap enough RIGHT NOW, regardless of the fact that I have a 500GB SATA drive in my system in order to store movies and images.

There are also valuable business applications for the same technology. If 64GB flash were "relatively affordable" and noticeably faster and more effective than a good RAID array, then these flash drives could be an important component of an enterprise storage system. Think "cache times ten, plus non volatile."

I'd love to build a Lustre Cluster where each of the nodes had one of these as a persistent backing store for cache: http://www.fusionio.com./ As an added plus, their stated market price of $30/GB is ALMOST worth it.

C//

Re:With the Exception of What???

[Bryan Ischo]

I have never used more than 20 GB on any system I have ever owned. I can go years and years on the same system without using significantly more space. But I don't pirate music and movies, and it's pretty hard to fill up a hard drive without doing those things (I guess if you install lots of software that could do it too, but the "working set" of software for most people is still probably pretty small and you can always uninstall stuff you are not using to make room for new software). I would be exceedingly happy with a fast, reliable, low-power, and cheap 32 GB solid state drive.

So, not everyone is like you, and needs constantly expanding amounts of persistent storage. For most people who are not pirating music and movies, modest hard drive capacities are just fine. Hell, even if you do pirate movies, you can always erase them after watching them instead of building up bigger and bigger collections of stolen movies.

Re:With the Exception of What???

[Bacon Bits]

So why not move to mag tape?

Re:With the Exception of What???

[chmod a x mojo]

sorry to say but not that many people are like YOU.
My computer has .8 TB storage and I need more. I don't pirate movies ( I have a service for converting home movies to professional menu based DVDs ) just try storing even 10 minutes of uncompressed video on your 20 GB drive and tell me how it goes :-|

not to mention the latest games taking anywhere from 2-10 GB to install, on top of VISTAs what ... 15 GB recomended ( don't use windows so I really don't know, but I think I remember hearing 15 for vista itself).

Re:With the Exception of What???

[Nom du Keyboard]

could make productive use of a 32 or 64GB flash system if it were fast enough and cheap enough RIGHT NOW, regardless of the fact that I have a 500GB SATA drive in my system in order to store movies and images.

And I'd use the extra money to buy a couple more big SATA drives and set-up a RAID 5 for better performance *and* data security.

Re:With the Exception of What???

[m50d]

Because having to change media is that little bit *too* inconvenient, and access is slow enough to be a problem. As the GP said, modern magnetic drives are fast enough; if you can play HD video and the latest FPS in realtime from them, what benefit will a home user see from upping the transfer speed? I've brought four hard drives in the past two years, and not one of them was because my current drive was too slow, or power consuming, or anything like that, but because my drives were full.

Re:With the Exception of What???

[Courageous]

That depends, doesn't it?

For example, this device here http://www.fusionio.com/ if affordable enough, would beat any RAID array you're ever likely to put in a workstation. In fact, it's hard to beat a device like this with ANY commercial RAID array. NetApp 3070's can't pull those numbers, and when they manage to get close, you're talking 14 15K FC drives.

Other "features" of your RAID array might be questionable as well. Like the power. And the noise.

One final factor. At SOME price point the fact that there is a ratio difference in price points becomes irrelevant. If you could get a 64GB fusionio card for 20 bucks, you wouldn't really care that a 64GB hard drive is 20 cents. The 100:1 difference in price is nothing, because both prices are nothing. At that scale and price point.

"Consumer price insensitivity" is the term. At some price, the price doesn't matter.

That day WILL arrive.

C//

Re:With the Exception of What???

[Bryan Ischo]

Are you saying that most people are like YOU, and have services for converting home movies to professional menu based DVDs? Somehow I doubt that. Most people who are filling up big hard drives are doing it by pirating movies and music. Period. Which is exactly my point - except for people who are pirating stuff, which honestly I don't think is a large percentage - maybe 5-10% tops? - don't need exceedingly large drives.

As to the latest games, how many latest games do people typically need to have installed at once? This is an honest question, I don't know. But I'll bet that once again the number of people who need to have multiple 10 GB games installed at one time is a pretty small fraction of the computer using population.

RAID: speed & reliability

[Anonymous Coward]

site details an early RAID experiment using 4 thumb drives:
http://www.bigbruin.com/reviews05/thumbraid_1

Hmmmm...thumb drives (USB flash drives) are:
1) hot swapable...plug-and-play
2) inexpensive...widely available...falling in price
3) daisy-chain USB hardware is expandable

Seems like there should be an idea for a product in all this.
RAID could solve the reliability issues...have the OS
pop open a warning box:
        Thumb drive #4 has now become 'unreliable'.
        It's data has been copied to other drives.
        Please replace #4 with a new device.

Is NV that important?

[igb]

There are also valuable business applications for the same technology. If 64GB flash were "relatively affordable" and noticeably faster and more effective than a good RAID array, then these flash drives could be an important component of an enterprise storage system.

We're going to see how effective this is over the coming months: NAS and SAN products are clearly going to start sprouting SSDs either instead of the primary cache or as a mid tier between the RAM and the disk. I'm not expecting miracles: RAM is very, very cheap and the cost mostly comes from the infrastructure around it. To take an example, my NAS storage has 12GB of mirrored RAM (you're still going to need to mirror the SSD unless you're insane) backed by enough battery to hold it up through a power failure until I can get enough power to stage it to disk. It's also using dedicated space within the RAID groups to use a an (essentially) linear-write intent log. An SSD might help a bit: perhaps 64GB of SSD is cheaper than 64GB of RAM and a battery. But at the moment for quite a heavy workload --- 2TB of Cyrus, 2TB of Clearcase, 6TB of Oracle, ~1000 home directories --- I never get to the position of writes not being staged via the RAM. Sure, I can imagine pathological workloads where I would need more staging, but right now 12GB is fine. So 64GB of SSD would add nothing to the proceedings except perhaps a small amount more read cache, and if the world were demanding more read cache (and it tends to exhibit diminishing returns) the answer would be to partition the RAM between battery-backed and mirrored write cache and unprotected read cache. ian

Re:Is NV that important?

[Courageous]

All good thoughts. Agree and all that. This being slashdot, instead of talking at length about agreement we'll have to up the ante and disagree about something (hah). Let's talk about that "you still need to mirror the SSD until your insane" comment. I'm finding in-box redundancy to be less and less useful. Check out http://www.isilon.com./ These guys, and emerging vendors like them, are beginning to move away from in-array replication of data to cross-array/cross-box replication. Packets are striped across the filers, instead of inside of the array. The beauty here, is that you can actually lose the filer chassis entirely, and you don't lose any data. You can replace when convenient, and get closer to the idea of "ignoring the chance of RAID-5/RAID-6 simultaneous disk failure". It's still not entirely there, mind, but since RAID rebuilds involve any disk in all of your racks, you have no need to do anything, the system is essentially self-healing.

Soon, Lustre file system will have such a feature. I can't wait.

Another vendor, mostly still in stealth mode, is here: http://www.xivstorage.com/

I'm sure you will find more and more turnkey cluster storage vendors hitting the market soon, precisely because clustered storage solves bandwidth, capacity, and manageability problems so well. All sorts of serviceability problems are cleanly solved by breaking away from classic RAID and moving to the right kind of cluster architecture.... problem is, until recently it's been a science project. Turnkey cluster appliances are right around the corner (or here, if you can live within Isilon's NAS-only constraints).

C//

Re:Is NV that important?

[igb]

It doesn't matter if the replication is within a chassis, within a closely coupled cluster or between boxes linked only by GigE: the point is that only a fool would take a write and place it into a single device, be that device RAM, SSD or rotating media. I currently use the classic ``two sets of electronics fronting one RAID array'' assemblage (Pillar) with replication taking place within a few hours to another box (a pile of second-user DotHill disk, a Sun, the magic of ZFS) twenty miles away. Except there I can do the replication faster than periodic (Oracle archive logs, Cyrus replication) I do. Hourly, daily and weekly snapshots at each end, periodic tape for long-term audit and I'm relative relaxed.

But you're right: this is all up for grabs in the coming months. Distributed filesystems are one thing, although the need for diffuse bandwidth covering the clients and all the nodes in the cluster could be a constraint for a lot of people. Companies like Acopia that use switches to virtualise NFS into mirrored milesystems are another (although as getting application vendors to accept NFS rather than SAN is a major hassle, virtualised NFS may cause them to tear up your support contract). Applications that are distribution aware are a third --- I hear from my Oracle administrators that Oracle.next has its own, userland NFS client so that it can play database-aware tunes on write sequencing taking advantage of NFS's ``a confirmed write is durable'' semantics, and it's an obvious move to just write your logs out to your remote filesystem as well as your local one.

I'd be interested to hear of people who are running Oracle into things like Lustre. It should work, and work well, but I suspect it will take a brave man to sign it off into production.

ian

Re:Is NV that important?

[Courageous]

I'd be interested to hear of people who are running Oracle into things like Lustre. It should work, and work well, but I suspect it will take a brave man to sign it off into production.

The problem with Lustre is that if you configure in the fashion that one usually uses Lustre, there is a small non-zero chance that something can happen to your cluster such that you can realize an availability problem that may impact the entire tier. A future version that does the distributed stripe thing will address that.

(Note that there are still good uses cases for Lustre regardless of the possible tier availability issue, and for the most part the problem can be addressed independently by having each Lustre node be itself HA; note however that in all standard HA patterns, total loss of the storage chassis still spells doom to recent non mirrored data... BTW, if you think storage chassis loss isn't necessary to account in your availability model, then you're golden... depends on what you're doing).

If you're interested in clustered storage ala Acopia, but need it on a SAN, talk to the people at XIV Storage. They're doing clustered storage for block-level data, and are able to easily keep ~40gbit/sec fed. Course, this is a startup. Magna caveat emptor.

C//

Reviewer has little reliability

[Anonymous Coward]

He talks about this "unknown" laptop seeming faster than the sony he had just tested, but doesnt tell us what this laptop is?

He also, in the sony review, says the sony is small because of this new SSD. That is just not true because the Sony has been on the market for a while with a standard hdd, and the SSD is just an upgrade.

Thus, I dont trust this reviewers very subjective opinions.

WANT

[Spyder]

The wear issues originally faced by this tech seem to have been addressed (2mil hours MTBF and at least 5x the sector rewrites), and they have become big enough to do most stuff especially with the current availability of external storage. With a nameplate like Samsung behind it I'm willing to give it a spin.... or um, try :)

Now has anyone found any place to GET ONE? I've been looking and I can't even find a model/part number. WTF? Why can't I be the first one on my block to have a 0 spindle laptop? It's a conspiracy I tell you.

"Up to"

[rrohbeck]

in a technical spec sheet? I smell something if they don't spec the minimum.

should last 140 years @50GB of writes/erasesperday

[danwat1234]

A SSD from the company, Mitron should last 140 years if you were to write/erase 50GB per day, every day. Seems like it should last 20+ years for a heavily used workstation/laptop. The main thing is to not run defrag at all, except to keep fragmentation levels OK, not that it really matters in an SSD. This information is from this link:: http://www.anandtech.com/storage/showdoc.aspx?i=3064&p=2 (on the left side of the table at the top of the page)

Mod parent insightful

[i.of.the.storm]

Very true, especially since SSDs cost about as much or more than that. The only reason you might have is that you need a 64 bit OS to handle 4GB of RAM. Due to addressing issues, even though a 32 bit OS can address 4GB of memory a lot of the addressing space goes to bits of hardware besides RAM, and graphics RAM takes a large chunk of that so most users will see around 2.8-3.3GB as their effective maximum RAM in 32 bit Windows. SLI/Crossfire kills that - imagine (just for sake of argument, this is completely pointless) having dual 1GB 8800GTs! However, I'd say Vista x64 is as compatible as it's going to get on the Windows side, so now's the time to jump on the 64 bit bandwagon. I have, and it's great (well, by great I mean as good as Windows usually is, which is great for me but maybe not for others) although I don't have 4GB of RAM yet.

Gigaseconds? Miliseconds?

[Anonymous Coward]

Samsung says the drive can withstand an operating shock of 1,500Gs at .5 miliseconds

Someone please teach them how to spell physical units.

"it makes virtually no noise when it operates"

[Anonymous Coward]

Virtually? Do Computerworld reporters hear the bits flipping?

but you didn't hear it from ME

[v1]

(Editor's note: We're honoring a request from Samsung Electronics Co., Ltd. not to reveal the make or model of the laptop the company supplied for our testing purposes.)

Apple MacBook Pro

where are your logs stored?

[goga_russian]

do you store the logs on CF? can you describe your setup a little better

Re:where are your logs stored?

[WindBourne]

Here is the hardware:
I used to hook up the IDE->CF. But the next time, I do this, I will use this instead (cheaper and does not take up a slot). In addition, absolutely do not use the cheap CF garbage. There is lots coming out of China and the quality is horrible. If you do use one of the cheap one and it goes bad, you will at least understand why quality costs. I used Sandisk. I bought it at micro center since it was close,but I would go with newegg if ordering off the web (lots cheaper).

As to the software, it is pretty much a standard install.
Install / to the CF. Keep it SMALL. I am using kubuntu these days, so they automatically do it small. During the install, I added /home from the disks, as well as /opt. Copy /etc to a disk. (for me, I back it up to /opt/backup/etc). After all this, I installed, apache, postfix, postgres, Mythtv (requiring Mysql), and squid. I elected to leave the postfix data on the CF (home server; small amounts of email; on the net, I use a gmail accounts, as well as nospamxxx accounts arriving at my system to avoid spam ).
I actually decided to leave the logs on the CF. They are the one thing that keep causing a disk to spin up.

I moved the data areas of apache, postgres, mysql, and parts of mythtv to the hard disk. They were located in /var. (back up /etc again)

Squid is in a tmpfs on the system. I figure that since I reboot infrequently, it may actually help to clear it.

BTW, I have a couple of gigs of ram in the server. I turned off swap. All in all, my disks now spend the vast majority of their time sleeping, powered down, with the server drawing very little power. Several have commented about the seasonal change, but I started measuring about 1 week after the re-build. The fact that the temp dropped so much will tell you that less power is being used.

Re:where are your logs stored?

[Unoti]

Wow, this was more interesting and thought-provoking than 90% of the articles I've read on Slashdot in the last week, and it's just a post!

Thanx; BTW, What amazes me

[WindBourne]

is that none of the major distros do a set-up like this. It should be relatively easy to do. In fact, for North America/EU/Oceanea(sp)/etc. a significant amount of our power goes to computers. If we can lower the use by creating a single home server, and then park the disks, I suspect that we could lower the total electricity by at least several percent.

Re:Thanx; BTW, What amazes me

[Anonymous Coward]

I eliminated a server a couple of years ago.

For the decade between 1996 annd 2006, I had been running a mostly-dedicated *nix box (sometimes Linux, sometimes FreeBSD) as a NAT router.

I recently "upgraded" that machine (most recently a P3-750) to a Linksys WRT54GS running OpenWRT. It still does all of the fancy routing that I'm accustomed to, and with the 512MB SD card I've hacked into it, it has a lot more storage space than the 486 router I was using 12 years ago.

It's also significantly faster, and uses something on the order of 10 Watts of power. Who needs a hard drive? Or a PC?

It seems likely that more tasks which are currently being performed by fast general purpose computers will be able to be miniaturized to this scale as time progresses.

For example, I used to spend some amount of time planning an in-car MP3-playing computer (just never had the cash to implement it), but these days most after-market car stereos handle MP3 just fine with relatively minimal energy consumption. I'd probably have never noticed the energy consumption of the MP3 PC+inverter were it ever to have seen the light of day (it's nothing compared to the energy required to push 2 tons of steel down the road), but still...

42KB!!!

[Vr6dub]

I get it dude! 42KB is all you've lost. You type of people drive me crazy at work.

Usefull in Laptops

[fast turtle]

I've been considering one of the 32GB laptop models because of power consumption. Hell I'd be damn happy to get an extra 3/4 hr battery life from the damn thing because that's more important to me then speed. As to large storage needs, that's what my desktop is for. Main thing is, battery life. More is always better and I'd like a laptop that runs for a minimum of 72 hours before the battery dies.

We're only at 1TB?

[sootman]

Why no love for the 5.25" form factor? That extra inch-and-three-quarters gives you a lot of extra real estate to play with. ((5.25/2)^2) / ((3.5/2)^2) = 2.25, if I'm doing this correctly, so even minus room for the spindle, etc., you're still talking about 5-100% more area.) Why is no one making a modern version of the Quantum Bigfoot* that came in my sister's 400 MHz Compaq Presario 5150? I would love to see a modern 5.25" HD with...
- 3600 or 4200 RPM rotational speed
- low noise
- low heat
- low power consumption
The reduced speed (wear and tear on parts) and heat should also lead to greater reliability. If a 3.5" drive can be 1 TB today, a 5.25" drive should be 1.5-2TB. A drive like this would be perfect for a home media server or HTPC, where performance is not critical (SD DVD is only 4 GB/hour; even BluRay is only 25 GB/hour--and I'm pretty happy with ripped DVDs at ~1500 kbps--less than 1 GB per hour) but low heat, low noise, and low power consumption are all desirable traits. (There's more rotating mass, but at lower speed there should be much less energy/momentum/intertia/whatever overall.) And as long as CDs and DVDs are still ~5"--and that seems to be the case (DVD, HD-DVD, BluRay, SACD)--we'll already be using properly-sized cases.

* granted, that old thing was slow as hell. Swapping out the stock 8 GB Quantum Bigfoot for a 30 GB Maxtor dropped boot times from 3 minutes to 40 seconds.

Re:We're only at 1TB?

[POTSandPANS]

I would think that you could still get pretty decent speed out of it. If you were to compare the speed of a 5.25" and 3.5" drive both running at 5400 rpm, the outside edge of the 5.25 will be moving past the heads quite a bit faster than the 3.5". Sure the access time wouldn't be as good, but for streaming media or a backup drive, I think the extra space would be worth a slightly lower access time.

been there, done that, the answer is NOT YET

[mauri]

I decided to test SSD drives on couple of laptop users some months ago.
Today we have none of them left, all went bad in a matter of weeks.
Tried SanDisk 5000 series, both 2.5" and 1,5". No luck.
1,8 died completely, 2.5 just got more and more bad blocks.
Will try with Mitron 7000 as well, when the damn thing ships.

But whatever they say, my suggestion is to keep out of this SSD business until there is more reliable NV memory than flash...

p.s. Writing is sloooowwww, I have commented it earlier here

I hope someone knowledgeable replies to your post

[FoamingToad]

I'd also like to see 5.25" large-capacity drives given another try, unless there is a specific reason not to.

More than once when upgrading other people's machines I've been forced to use 3.5" -> 5.25" drive rails in order to get the cable placement just right, and these days a modern tower case will typically give you three or four 5.25" bays - after throwing a dual layer DVD+RW what else realistically needs those bays?

I'd like to see a largish-capacity 5.25" drive - not necessarily all that fast, as long as it's capable of transferring enough data to saturate gigabit ethernet (on my server -> workstation setup I get about 30% throughput, max). Say a terabyte, maybe in two platters, at perhaps 4200 revs. Any takers?

Re:I hope someone knowledgeable replies to your po

[Alt_Cognito]

I know this is late in the posting of this article, but the main reason is manufacturing costs. The cost of building a factory to manufacture 5 1/4 plates probably does not meet the profit margins of simply maintaining a single manufacturing process. Less engineering, less factories etc... The more homogeneous their factories are, the easier they are to upgrade etc...

Interesting; some questions for you.

[WindBourne]

In my house server, it acts as firewall, router, mythtv, nfs, asterisk, postfix, postgres, even game server, etc. Basically, one box covers everything. I have been playing with XEN on one of my boxes and am thinking that I will switch to that on the server. One machine for the outside world, and several for the inside.

Is the WRT54GS capable of that? IOW, can I plug in a tv card (slot or USB)? In addition, it would be useful to have it serve the asterix up.

If I can lower the energy usage for my systems inexpensively, I would gladly do it.

Re:I hope someone knowledgeable replies to your po

[sootman]

Good point. That may very well be the reason. But there's GOT to be a market for drives with 2x the capacity of anyone else's! And hard drives typically don't follow a straight bigger-equals-better-value curve--there's usually a sweet spot in the middle (like where 320s/500s/maybe 750s are today) and then an upswing at the high end (where 1 TB is today.) Just like with LCD panels--medium is a better value than small, but large is a worse value than medium. So people wouldn't mind if a 2 TB drive cost more than 2x a 1 TB drive.