Beta
×

Welcome to the Slashdot Beta site -- learn more here. Use the link in the footer or click here to return to the Classic version of Slashdot.

Thank you!

Before you choose to head back to the Classic look of the site, we'd appreciate it if you share your thoughts on the Beta; your feedback is what drives our ongoing development.

Beta is different and we value you taking the time to try it out. Please take a look at the changes we've made in Beta and  learn more about it. Thanks for reading, and for making the site better!

OCZ RevoDrive 350 PCIe SSD Hits 1.8GB/sec With Standard Toshiba MLC NAND

timothy posted about 4 months ago | from the and-this-time-next-year dept.

Data Storage 113

MojoKid (1002251) writes "OCZ was recently acquired by Toshiba and has been going through its product stack, revamping its SSD portfolio with fresh re-designs based on Toshiba NAND Flash memory for not only increased performance but better cost structure as well. OCZ has now replaced their RevoDrive family of PCIe SSD cards with an almost complete re-designed of the product. The RevoDrive 350 is based on the same OCZ VCA 2.0 (Virtualized Controller Architecture) technology as the previous generation but is now enabled with a PCI Express X8 card interface and up to 4 LSI SandForce SD-2282 SSD processors, along with 19nm Toshiba NAND Flash. The good news is, not only is the new RevoDrive 350 faster at 1.8GB/sec claimed bandwidth for sequential reads and 1.7GB/sec for sequential writes, but it's also significantly more affordable, at literally half the price of the previous gen RevoDrive 3 when it first launched. In the benchmarks, the new PCIe card excels at read throughput, regularly hitting its 1.8GB/sec claimed bandwidth, especially with sequential workloads. Write performance is solid as well and the drive competes with the likes of some higher-end and more expensive SLC NAND-based PCIe cards like LSI's WarpDrive and Intel's SSD 910."

cancel ×

113 comments

Sorry! There are no comments related to the filter you selected.

Slashdot Beta is driving up my blood pressure (0, Insightful)

Anonymous Coward | about 4 months ago | (#47012389)

Down with the Beta! Just toss it out and be done with it, Dice.

Re:Slashdot Beta is driving up my blood pressure (0)

Anonymous Coward | about 4 months ago | (#47012575)

Try this link. [slashdot.org] Same story, better formatting.

No. Absolutely not. (3, Interesting)

stonecypher (118140) | about 4 months ago | (#47012405)

My RevoDrive failed in three weeks of light use, and they refused to honor the warranty.

Toshiba also refuses to honor the warranty, despite that they admit that the purchase was real and that the existing warranty was not honored, and despite that I am a standing Toshiba customer.

Therefore nothing has changed, and you do not want a RevoDrive.

Re:No. Absolutely not. (2, Interesting)

Anonymous Coward | about 4 months ago | (#47012485)

What was the basis of their refusal?

Re:No. Absolutely not. (3, Informative)

Anonymous Coward | about 4 months ago | (#47012695)

He putted it in his microwave.

Re:No. Absolutely not. (1)

Richy_T (111409) | about 4 months ago | (#47012941)

How else are you supposed to dry the beer off?

Re:No. Absolutely not. (2)

click2005 (921437) | about 4 months ago | (#47012959)

On your barbeque obviously, the beer helps prevent cancer.

Re:No. Absolutely not. (1)

davester666 (731373) | about 4 months ago | (#47015321)

You can get cancer from SSDs? I learn something new every day!

Re:No. Absolutely not. (0)

Anonymous Coward | about 4 months ago | (#47013055)

That's a big microwave. Tell me, was he using a regular putter, or one of those standing putters that the USGA doesn't allow anymore? Also, was his caddie in there with him?

Re:No. Absolutely not. (0)

stonecypher (118140) | about 4 months ago | (#47014101)

There was none. They simply refused.

Re:No. Absolutely not. (5, Informative)

NormHome (99305) | about 4 months ago | (#47012657)

Unless the unit was damaged, modified or some grey market edition I don't know that they have a legal leg to stand on, you can file a complaint with the state attorneys generals office allege fraud for failing to honor the warranty. It also depends on what state you're in i.e. "Void where prohibited" since some states have stronger consumer protection laws than others and some of the warranty terms may not apply to you.

Why did they deny your request for a warranty replacement?

Re:No. Absolutely not. (2)

Guspaz (556486) | about 4 months ago | (#47013185)

OCZ probably had no legal leg to stand on, but Toshiba is not legally required to honour OCZ's warranties, even though they bought the company.

Re:No. Absolutely not. (1)

riis138 (3020505) | about 4 months ago | (#47013663)

I find it odd that Toshiba would not honor the warranty on good faith. I have always had very good experiences with them in the past being an owner of multiple Toshiba laptops.

Re:No. Absolutely not. (4, Informative)

Guspaz (556486) | about 4 months ago | (#47013821)

I, on the other hand, have had terrible experience with Toshiba's US support department.

I live in Canada, and wanted to buy a model of Toshiba laptop that was only available in the US. Their warranty claimed to be international, and I called Toshiba USA to confirm this. At worst, I was told, I'd have to pay the shipping costs myself. I was fine with that.

Then I dropped the laptop. It was fine electronically, it was just that a chunk was missing from the chassis, so I needed to replace some parts of the laptop chassis.

I called Toshiba USA. They refused to have anything to do with me, saying that not only could they not ship the laptop back to me at my expense, they couldn't even RECEIVE the laptop at their repair centre. Their shipping department would refuse to accept any shipment from outside the country. They told me my only solution was to mail my laptop to somebody in the US and have *THEM* send it to Toshiba. Mind you this laptop was only a few months old, only a few months into a 3-year warranty.

I called Toshiba Canada. They refused to repair the laptop (at my cost) even though by then they sold the same laptop in Canada as my American model. They said that because the model numbers were different (even if the laptops were identical) they wouldn't touch it.

Because of this, the only way that I could get my laptop repaired was to mail it to one of Toshiba's authorized third-party repair companies in the US. In the end, I had to pay $600 to repair a $1200 laptop... even though only the chassis (which probably costs a few dollars at most) was broken, with all electronics perfectly fine.

To rub salt in the wound, they did a terrible repair job, violating Toshiba's own service manual, and incorrectly re-installed the keyboard.

As a result, I will never buy another Toshiba product, nor will I ever recommend anyone buy one. Considering that many of my friends have already been burned by OCZ, them being bought by Toshiba changes my recommendation to "Don't buy OCZ products because they have a super high failure rate and Toshiba will try to screw you over on replacements."

Re:No. Absolutely not. (2)

AmiMoJo (196126) | about 4 months ago | (#47015833)

On the other hand I have had excellent service from Toshiba. My TV wouldn't take a firmware update so they collected it, fixed it and returned it quickly and at no cost to me. It was about 18 months old at the time, no problem with the mandatory 2 year warranty that some companies like to quibble over.

Years ago when I used to fix laptops for a living they were good on parts and warranty repairs too. It seems like your slightly unusual situation is what caused problems. Of course that is no help to you, but it doesn't necessarily mean that for someone who buys products in their home country there is any kind of problem. I'm sure you can find cases that all companies fail to handle well.

Only time will tell if the new Toshiba SSDs will be any good, but I think I'll wait for real stats instead of just going by a few Slashdot anecdotes.

Re:No. Absolutely not. (1)

Crookdotter (1297179) | about 4 months ago | (#47016527)

When given the option of $600, why didn't you just fix it with some epoxy or somesuch?

Re:No. Absolutely not. (1)

Guspaz (556486) | about 4 months ago | (#47016745)

The chassis of the laptop was a magnesium alloy, and "shattered" was a better description than "cracked". It was beyond the ability of repair with the existing parts, it needed replacements.

Re:No. Absolutely not. (2)

stonecypher (118140) | about 4 months ago | (#47014109)

Toshiba is not legally required to honour OCZ's warranties

Who cares? If you buy a company that performs fraud, admit that the fraud happened, and refuse to make right, it is legitimate for the victim to warn other people away.

Re:No. Absolutely not. (1)

beelsebob (529313) | about 4 months ago | (#47014679)

Actually, they are legally required to honour it. Part of buying the company is that you end up in all the contracts that the old company was in, and you absolutely must honour them. You can't get out of contracts you have with people simply by setting up a new company, and buying your old one out with it.

Re:No. Absolutely not. (2)

Guspaz (556486) | about 4 months ago | (#47015019)

Warranty holders are creditors. OCZ Technology Group went bankrupt, and Toshiba bought OCZ's assets after that. Toshiba then launched a new company named OCZ Storage Solutions.

In other words, OCZ's assets were liquidated, and some creditors got paid. After that, OCZ the original company didn't exist anymore, and the warranties were worthless. Toshiba theoretically decided to honour some (but not all) warranties, but they were not obligated to do so.

Re:No. Absolutely not. (1)

drinkypoo (153816) | about 4 months ago | (#47017631)

OCZ probably had no legal leg to stand on, but Toshiba is not legally required to honour OCZ's warranties, even though they bought the company.

Sometimes that sort of thing is a condition of a merger. Remember, corporations are supposed to function in the public interest. Are you in fact sure that Toshiba is not legally required to honor OCZ's warranties?

Re:No. Absolutely not. (1)

stonecypher (118140) | about 4 months ago | (#47014107)

you can file a complaint with the state attorneys generals office allege fraud

Not worth my time. It's far more cathartic, besides, to steer tens of thousands of potential customers away with public recitation.

Re:No. Absolutely not. (0)

Anonymous Coward | about 4 months ago | (#47017663)

Another option is your credit card company. Some will try to help you fix it. Mine has something like 1 year extra on purchases over 250 I think.

As I tell my friends 'my credit card company are bastards but they are my bastards'.

Re:No. Absolutely not. (2)

Sasayaki (1096761) | about 4 months ago | (#47014541)

In Australia this would never fly, ever.

The ACCC a few years back put in a new law (which Apple fought tooth and nail, source: http://www.afr.com/p/technolog... [afr.com] ) which required every piece of electronics sold in Australia to have a two year "warranty". I put that in sarcasm quotes not because it's invalid (the ACCC has some *serious* bite here, enough to scare Apple into compliance), but because it's not technically a warranty. It's simply: "a reasonable expectation that an electronic product will be fit for purpose for two years from purchase".

Legally, that's not a warranty, but in some ways it's a lot more powerful.

However, Apple continues to fight it, usually by simply redefining their terms. For example, I had a 1.5 year old iPhone 5's battery die recently. I took it in to get replaced, they said that batteries are considered consumable items and, based on its charge/discharge cycles, it had been "consumed", rather than "broken" or "worn out".

I went home, printed out the relevant law, returned and showed it to them and the manager replaced it for free, all the while warning me that this wasn't something they were expected to cover. The girl helping me was very sympathetic and helpful, though, and I felt as though both the manager and the genius-bar chick both resented Apple dodging the law a little bit.

If your RevoDrive failed in any way for two years after purchase here, in Oz, it would get fixed for free. Not even Apple can dodge that.

Re:No. Absolutely not. (0)

Anonymous Coward | about 4 months ago | (#47015129)

In Australia this would never fly, ever.

The ACCC a few years back put in a new law (which Apple fought tooth and nail, source: http://www.afr.com/p/technolog... [afr.com] ) which required every piece of electronics sold in Australia to have a two year "warranty". I put that in sarcasm quotes not because it's invalid (the ACCC has some *serious* bite here, enough to scare Apple into compliance), but because it's not technically a warranty. It's simply: "a reasonable expectation that an electronic product will be fit for purpose for two years from purchase".

When you dig into the Australian law:
"Though it does not specify how long a warranty should last for specific products, it must allow repairs within a reasonable time of purchase."
So for smartphones regularly sold on a 2 year contract, 2 years should be the minimum warranty period. Whitegoods are typically expected to last 5 years at least, so if your fridge fails after 3 you have the law on your side if the manufacturer only includes a 2 year warranty. There is however another expectation - spending a large sum on an item vs buying a cheap version of a similar item may affect the reasonable warranty period. Brand names on your goods increase the price, but also increase the expected lifetime (quality) of an item, which also increases the expected warranty period.

What about reliability? (4, Interesting)

jandrese (485) | about 4 months ago | (#47012429)

OCZ always struggles with reliability, and buying their Lambo performance hardware always seems like a recipe for lost data. The fact that they're pushing MLC flash chips to the limit is not reassuring.

Re:What about reliability? (3, Interesting)

AaronW (33736) | about 4 months ago | (#47012671)

Out of three OCZ drives I have had two failures within the first two weeks of use.

Last Friday my 2-week old work machine with an "enterprise class" OCZ hard drive hooked up to a Dell server with an LSI raid controller suddenly started corrupting the EXT4 filesystem all to hell rendering the machine unbootable. I had days worth of work on that machine. Fortunately I was able to recover the data from lost+found after fsck.ext4 but the root directory was totally lost along with who knows what other data.

At home I have two OCZ hard drives. One of them turned into a brick after two weeks of use. I got a replacement from OCZ and make sure I do an XFS dump every night onto my RAID drives.

Re:What about reliability? (2)

Hadlock (143607) | about 4 months ago | (#47012715)

That sounds about par for the course. Before the brand was purchased, they saw a 5% return rate across all of their drives, as high as 15% for specific models. What junk. I was shocked when I heard that they had been bought and rolled in to another company. It's going to be hard to overcome that kind of terrible reputation. There's a reason why no OEMs include OCZ drives in their offerings.

Re:What about reliability? (1)

Jeff Flanagan (2981883) | about 4 months ago | (#47013217)

>Out of three OCZ drives I have had two failures within the first two weeks of use.

Out of six OCZ SSDs, I've never had a single failure. They've only gone out of service when replaced by larger drives as they became affordable. They seem to be fine in my environment, but I'm not pulling power cords out while the PC is on.

Re:What about reliability? (0)

Luckyo (1726890) | about 4 months ago | (#47013629)

That makes one of you. Probably in the whole world.

Re:What about reliability? (1)

petermgreen (876956) | about 4 months ago | (#47014685)

Not really

Another poster claims that ocz had an average return rate of 5%. Some people don't return their drives because they value their privacy more than the cost of the drives or because they can't be bothered with the effort of returning them and some drives fail after the warranty runs out but before they are too small to be useful so lets suppose the overall proportion of drives has a 20% rate of "failure before becoming too small to be usable". Lets also assume that failures are independent events.

If my quick calculations are correct that would make the chance of exactly two out of three drives failing about 10%, it would make the chance of exactly zero drives failling out of 6 about 27%.

So with realistic guesses about failure rates it's very likely that there are a lot of people out there who have had multiple OCZ drives die and also a lot of people out there who have owned a moderate number of OCZ drives and had none of them die.

Re:What about reliability? (1)

ooshna (1654125) | about 4 months ago | (#47015119)

Yeah lets assume that 3/4ths of failures go unreported on drives that most people installed themselves. Its not like these are drives that came in their Dells or Acers. Most of these drives were installed by the people who own the computer or someone in a IT department. I doubt these 3/4th of these kinda people would no go to the trouble to replaces their expensive (compared to a HDD of much greater capacity they could have got) of equipment.

Re:What about reliability? (0)

Anonymous Coward | about 4 months ago | (#47015247)

5% AFR of over 5 years = 22% failure rate (ballpark, assuming linear failure rate etc blah blah).
A 80GB X25-M still makes a half-decent boot drive today.

So 3/4 of people don't want a refund? (2)

dbIII (701233) | about 4 months ago | (#47015367)

Ah, the silent majority argument which normally only pollutes political discussions. It falls down when you consider that it should also apply to spinning storage.
As for the additional numerology on a flawed premise - it's depressing to watch.

Re:So 3/4 of people don't want a refund? (0)

Anonymous Coward | about 4 months ago | (#47015525)

Yes, because annual failure rate calculated from warranty returns over the first 24 months == total failure rate over the 5+ year useful lifespan of a client SSD.
*facepalm*

Re:What about reliability? (1)

pahles (701275) | about 4 months ago | (#47015495)

Nope. Ok, for me it is one out of one, but no failure here, working just fine.

Re:What about reliability? (2)

eulernet (1132389) | about 4 months ago | (#47013265)

The bright side is that these new RevoDrives will corrupt the data a lot faster than before, so you'll notice the problem sooner ;-)

Re:What about reliability? (2)

LordLimecat (1103839) | about 4 months ago | (#47013603)

Sounds like you'd be better off not worrying about RAID and worrying about an actual backup system instead.

Re:What about reliability? (0)

Anonymous Coward | about 4 months ago | (#47014925)

Sounds more like a Linux bug to me, specifically ext4. It's been well known that ext4 has had a number of corruption bugs which may or may not have been fixed in the kernel version used on your server. If you were using Windows you would have used NTFS which is actually a very strongly tested filesystem (which you'd expect given its use in such a widely used environment). It's actually one of the few computer filesystems which you can find accreditation for in terms of being suitably valid to use in critical industries, whereas ext4 has no such perks because it was made by neckbeards.

Seriously, just use Windows and give up on that hobby crap.

Re:What about reliability? (1)

gshegosh (1587463) | about 4 months ago | (#47012977)

So it seems they basically obtain their speed levels by redirecting writes to /dev/null and/or /dev/random? ;-)

Re:What about reliability? (4, Informative)

tlhIngan (30335) | about 4 months ago | (#47013101)

OCZ always struggles with reliability, and buying their Lambo performance hardware always seems like a recipe for lost data. The fact that they're pushing MLC flash chips to the limit is not reassuring.

Except you're assigning blame in the wrong place.

It is not MLC flash that's the problem. It's OCZ.

The problem with OCZ is they don't put data reliability first - they put speed first. So they compromised reliability for speed. One famous example is where the SandForce controller could run in two modes - a safe-but-slower mode where you don't need external power caps, and a fast-but-unsafe mode where you need power caps so it can dump the tables to flash on power off.

What did OCZ do? They did the fast mode without power caps. So remove power suddenly and the FTL tables get corrupted, losing your data. It's fast, though, and cheap (since you don't pay $$$ for caps), but boy is it a ticking time bomb.

Considering SATA-III limits are hit (or why every SSD is now the same speed), there's no need for performance - using that extra speed for safety is far better.

(It's also why we have PCIe SSDs, because SATA-III is now too slow, and it didn't take long for SSDs, once the standard was reached, to hit the max).

Re:What about reliability? (0)

Anonymous Coward | about 4 months ago | (#47017405)

I have to factory reset my OCZ SSD every 6 months because it will start crawling at a snails pace. Other than that it works.

Too bad Samsung's XP941 is 2/3 the price (2)

Fallen Kell (165468) | about 4 months ago | (#47012519)

Seriously the XP941 is a native PCIe controller, not multiple SATA controllers raided together with a PCIe bridge controller. As a result, it is almost 1/2 the price, and still has similar performance (it is only a PCIe 1x device that does 1.2GBs reads/writes, vs the PCIe 4x device that only does 1.8GBs).

Re:Too bad Samsung's XP941 is 2/3 the price (1)

Tumbleweed (3706) | about 4 months ago | (#47012667)

Seriously the XP941 is a native PCIe controller, not multiple SATA controllers raided together with a PCIe bridge controller. As a result, it is almost 1/2 the price, and still has similar performance (it is only a PCIe 1x device that does 1.2GBs reads/writes, vs the PCIe 4x device that only does 1.8GBs).

Pretty sure that it's a PCIe 2.0 x4 device. :)

Re:Too bad Samsung's XP941 is 2/3 the price (1)

Gareth Iwan Fairclough (2831535) | about 4 months ago | (#47012857)

Seriously the XP941 is a native PCIe controller, not multiple SATA controllers raided together with a PCIe bridge controller. As a result, it is almost 1/2 the price, and still has similar performance (it is only a PCIe 1x device that does 1.2GBs reads/writes, vs the PCIe 4x device that only does 1.8GBs).

Pretty sure that it's a PCIe 2.0 x4 device. :)

To quote Mr Cole, "You can get more of what you want with a kind word and a bit of 2x4 than you can with just a kind word." B5.

Re:Too bad Samsung's XP941 is 2/3 the price (0)

Anonymous Coward | about 4 months ago | (#47015355)

Wonder how long until someone sticks 4 or 8 PCIe M.2 drives behind a PLX switch and writes a fakeraid BIOS+driver...

Still using Sandforce? (2)

snarfies (115214) | about 4 months ago | (#47012557)

NOPE. Not interested.

Re:Still using Sandforce? (1)

Bryan Ischo (893) | about 4 months ago | (#47012813)

Why? What's wrong with Sandforce? Works fine for me ...

Re:Still using Sandforce? (1)

Guspaz (556486) | about 4 months ago | (#47013201)

Works fine in Intel drives, but not so much in OCZ drives where OCZ has disabled all the safety features to eek out a bit more performance.

Re:Still using Sandforce? (1)

Bryan Ischo (893) | about 4 months ago | (#47014803)

Ah, so that's what he meant. Yeah I've had two Intel 520s in my work computer for a couple of years now, no problems. But you're right, the crucial difference is Intel vs. OCZ.

You can count me among those who wouldn't touch anything by OCZ with a 10 foot pole. Not even with Toshiba backing them.

Re:Still using Sandforce? (1)

dave420 (699308) | about 4 months ago | (#47016027)

*eke :)

Re:Still using Sandforce? (1)

thegarbz (1787294) | about 4 months ago | (#47014995)

Are you trying to say that OCZ won't screw up with a different controller?

$529 for the smallest (1)

oic0 (1864384) | about 4 months ago | (#47012565)

Sounds awesome but its so expensive I would rather spend the money on other parts for a more tangible gain.

Re:$529 for the smallest (1)

beelsebob (529313) | about 4 months ago | (#47014691)

Not that I'd recommend buying this SSD, but an SSD is in 99% of cases the largest tangible gain you can get in a computer. The fact that random reads (what slows down your system the vast amount of time) are 3 orders of magnitude faster than with a hard drive makes that the case in almost all cases.

Compilation: My Core i7 used only 2 cores while compiling reading off my HDD, it gets pegged to all 8 now, and compiles much much faster now.
Booting: Not that you do it often, but booting a machine is near instant with an SSD, including all applications loading.
Loading games: loading screens in games basically disappear with an SSD, or in the most intensive ones become very short.
Any time swap gets hit: Not that hitting swap doesn't slow your system down, it becomes tolerable with an SSD. It's rare, but it does happen occasionally, and it no longer kills your system.

Seriously, an SSD was the biggest gain in system performance I've seen in years.

Re:$529 for the smallest (1)

rudolfel (700883) | about 4 months ago | (#47015511)

Seriously, an SSD was the biggest gain in system performance I've seen in years.

It should also read "Seriously, an SSD was the biggest gain in data loss I've seen in years."

Maybe OCZ will improve under Toshiba (1)

maugle (1369813) | about 4 months ago | (#47012579)

Hopefully Toshiba will work to fix OCZ's infamously poor SSD reliability. 1.8GB/sec transfers mean nothing if the drive is so spotty you may as well be moving your files to /dev/null

Re:Maybe OCZ will improve under Toshiba (1)

viperidaenz (2515578) | about 4 months ago | (#47013169)

They'll just fix it by making it faster than /dev/null

Is this an ad? (0)

Anonymous Coward | about 4 months ago | (#47012581)

OCZ is notoriously bad. Why should these be any better? Only an ad would ignore their long history of DOAs and short-lived products.

Avoid OCZ at all cost! (2)

AaronW (33736) | about 4 months ago | (#47012607)

Last week I lost an entire day due to an enterprise class OCZ drive that decided to corrupt itself. My Linux box (using EXT4) started suddenly developing all sorts of filesystem errors rendering the system unbootable. The machine was two weeks old and had been used for software development. I also had another OCZ drive suddenly fail after two weeks where the drive turned into a brick. Reading up online on the new drive returned many reports of corrupt data from an "enterprise" class drive.

Who cares how fast you can access the drive if the drive can turn into /dev/random or /dev/null without warning.

Re:Avoid OCZ at all cost! (1)

Guspaz (556486) | about 4 months ago | (#47013225)

Sounds like a case of "should have used ZFS/BTRFS" on top of the case of "should have used anything but OCZ" :P

Re:Avoid OCZ at all cost! (1)

greg1104 (461138) | about 4 months ago | (#47015161)

ZFS will also corrupt itself in situations where the drive lies about writes. Running ZFS with unreliable writes has the same properties as running without NVRAM-protected storage [oracle.com] , which "can lead to data loss, application level corruption, or even pool corruption".

Re:Avoid OCZ at all cost! (1)

Guspaz (556486) | about 4 months ago | (#47015233)

If one of your drives is lying about writes, but the others aren't, ZFS should still be fine. When you try to read back the block, the data from the "evil" drive will fail the checksum, and the data will be reconstructed from the other drives.

Re:Avoid OCZ at all cost! (0)

Anonymous Coward | about 4 months ago | (#47015203)

Which would have helped... how?
Unless you have a RAID-Z or whatever the equivalent for BTRFS is or get *really* lucky, the only thing it can do when your drive decides to swap blocks around, replace them with older versions and/or zero them out (all common observed failure modes when FTL tables get corrupted) is to tell you that it's well and truly hosed.

Re:Avoid OCZ at all cost! (1)

Guspaz (556486) | about 4 months ago | (#47015227)

Ditto blocks can provide redundancy in a single-drive setup, but why would anybody run a single-drive setup for anything important these days?

Re:Avoid OCZ at all cost! (0)

Anonymous Coward | about 4 months ago | (#47015715)

Maybe it wasn't anything important? If you only lose 1 day of work it isn't really important, but that doesn't mean you want to lose it because OCZ produces crap.

Re:Avoid OCZ at all cost! (1)

GigaplexNZ (1233886) | about 4 months ago | (#47015433)

Who the hell uses Btrfs on production systems? It's not ready.

Re:Avoid OCZ at all cost! (1)

Guspaz (556486) | about 4 months ago | (#47015467)

Well, I would have just said ZFS, but while ZFS on Linux is pretty stable these days, ZFS boot on Linux isn't. And since the OP mentioned that it was his boot drive...

Personally I just slapped some old Intel SSD running ext3 in my server as the boot drive, because if that gets hosed, I'll just re-import the zpool on a new installation.

Re:Avoid OCZ at all cost! (1)

jon3k (691256) | about 4 months ago | (#47013851)

I have three OCZ drives (including an original 30GB OCZ Vertex) and all of them are still performing flawlessly.

Ignore these naysayers (1, Informative)

TheSimkin (639033) | about 4 months ago | (#47012655)

I have used over 16 OCZ drives and pushed them to their limits with no problems. Obviously all drives will eventually fail regardless of their nature, and you should always backup and use raid if uptime is important. Anyone using any drive without redundancy and backups should not affect your opinion of these drives.

Re:Ignore these naysayers (1)

Anonymous Coward | about 4 months ago | (#47012795)

One Clueless Zealot Obviously Cares Zilch Of Corrupted Zettabytes. OCZ's Credibility's Zonked.

Re:Ignore these naysayers (4, Informative)

Miamicanes (730264) | about 4 months ago | (#47013001)

The problem with many of OCZ's drives (like the Vertex2 and Agility2) was that traditional RAID wouldn't save you because whatever killed drive #1 could (and often did) kill one or more of your OTHER drives, too.

The fault lies 100% with OCZ. They omitted the supercapacitor that Sandforce's engineers intended to keep it powered up if it unexpectedly lost power during a write, and they compiled their drivers to NOT use the multi-step write strategy that a drive without backup power needed in order to write safely and recover gracefully from power loss (because multi-step writes killed performance).

Re:Ignore these naysayers (1)

petermgreen (876956) | about 4 months ago | (#47014721)

The other big problem with traditional raid is it only saves you if the drive either dies or reports a sector as unreadable. If it just quietly returns the wrong data then traditional raid doesn't help. That is one of the big attractions of filesystems like zfs and btrfs that combine checksuming with redundancy at the same layer.

Re:Ignore these naysayers (0)

im_thatoneguy (819432) | about 4 months ago | (#47014753)

It sounds like it's an incompatibility with Ext4.

I've got 10 drives and haven't had any trouble with mine or anyone I know. But here we have at least 4 instances on slashdot of Ext4 failures. Maybe the controller freaks out with Ext4 somehow.

Re:Ignore these naysayers (1)

greg1104 (461138) | about 4 months ago | (#47015135)

No, it's OCZ. ext4 is the most popular filesystem that expects good behavior from drive write caches, so of course it also has the most problem reports. The way write barriers work in ext4, the filesystem struggles when hardware lies about data being flushed to disk. See ext4 and data loss [lwn.net] for an introduction.

As outlined there, ext3 gets lucky in some situations ext4 just doesn't tolerate so some people see that as a bug in ext4. But the reason for the change is improved performance. You just can't get a fast filesystem and rugged behavior in the face of drives lying at the same time. You have to pick a side there. In the classic "good-fast-cheap--pick two" trio of trade-offs, OCZ always picks cheap and fast.

Bad drives aren't tolerated by zfs or btrfs either. It's just the case that ext4 is deployed on far more servers than they are.

Re:Ignore these naysayers (0)

Anonymous Coward | about 4 months ago | (#47013103)

Anyone using any drive without redundancy and backups should not affect your opinion of these drives.

Not so. If something has a high enough failure rate, then you start to notice that you're spending a lot of YOUR TIME on it. It doesn't matter that you can just pull the SSD from its RAID1 and put another one in its place and be resynced in a few minutes. What matters is that you are doing that every fucking weekend, instead of about once per year like you do with the lowest-priced "shitty" disks. Without moving components, SSDs are expected to be more reliable than thirty-dollars-per-terabyte mechanical storage, not less reliable.

Wake me up when OCZ's junk sells for less than thirty dollars per terabyte, because maybe then, I will be able to tolerate their substantially higher-than-rotating-disks failure rate. Until then, I want OCZ's time-wasting crappy products to be someone else's problem, not mine.

Obsolete tech. Pass. (1)

Anonymous Coward | about 4 months ago | (#47012677)

This is just a bunch of SSD's on a card with a raid controller chip. It's an old tech that's already obsolete.

SATA6bps is the bottleneck and it's already been eliminated. The future is PCI express native SSDs and they've been here for more than a year in laptops (Notably apple ones)

Just this week Intel released the 9 series chipsets that support both SATAExpress and the M.2 SSD format. Bot provide 4 lanes of native PCI Express.
http://en.wikipedia.org/wiki/M.2

AHCI based devices are out now. (Most modern OSs support AHCI natively) NVMe based devices will be out soon and will be even faster, with a porotocol that wasn't originally designed for magnetic hard drives.

Funny how tech seems to always come full circle. Does anyone remember what IDE meant? Integrated Drive Electronics - The original IDE drive interface was nothing more than an extension of the ISA bus, the primary system device interconnect, over a 40 pin ribbon cable. What are we doing now? Extending PCI Express, the primary system device interconnect, over thin little cables.

intel's DMI bus is still to slow (1)

Joe_Dragon (2206452) | about 4 months ago | (#47012919)

They need to have a better DMI bus even the high end cpus with 40 pci-e 3.0 lanes use as well.

When you have a few pci-e 2.0 lanes + sound + network + firewire + usb + sata + other IO all on the that dmi bus that is only pci-e 2.0 X4 you are not going to get all that you can out of an pci-e ssd.

And the pci-e 3.0 lanes need to be more in the low-mid-to lower high end cpu's desktop 16 3.0 lanes some times spit to X8 X8 is to Little.

Some boards like to put usb 3.0 + TB and other stuff on the X16 line takeing it from the video card.

Re:intel's DMI bus is still to slow (0)

Anonymous Coward | about 4 months ago | (#47013083)

You're not going to saturate the DMI link with you gaming computer's SSD. (But you do have a point. There is only one SATAExpress port for a reason)

The high end 9 series chipsets are for gaming, high-business desktop use.

If you need 30 lanes then you go to the it's-really-a-xeon high-end-gaming/workstation platform with it's quad channel memory controllers and bells-n-whistles. That platform is going to be refreshed later this year.

Re:intel's DMI bus is still to slow (0)

Joe_Dragon (2206452) | about 4 months ago | (#47013165)

missing the point.

Both the xeon high-end-gaming/workstation platform and the low to high end 9 series chipsets are for gaming, high-business desktop use. Put USB, networking, chipset IO, the few X1 slots all on the DMI bus.

Now maybe if there was a board that gave video X16 2.0 / X8 3.0 and used the other X8 3.0 for some pci-e sdd slots / maybe 1-2 pci-e X4 2.0 slots.

at least with the high end board you can have 2 full X16 3.0 slots + 8 3.0 leftover (per CPU). So them they can put pci-e 3.0 / switched 2.0 from the cpu or put network, usb and so on the it's own and not on the slow DMI bus.

Re:intel's DMI bus is still to slow (1)

Anonymous Coward | about 4 months ago | (#47013447)

The DMI link is 20 gigabits. 20 fucking gigabits and the graphics card isn't even on it. The DMI link is not fucking "slow"

  I really, really, really doubt there's anything that the intended target audience of the platform that will every come close to needing 20 gigabits.

If you are doing something that saturates 20 gigabits then you need to pony up for a more expensive platform. End of story.

Re:intel's DMI bus is still to slow (0)

Joe_Dragon (2206452) | about 4 months ago | (#47013771)

DMI bus is only pci-e 2.0 X4.

each network port is pci-e 2.0 X1
usb 3.0 is pci-e 2.0 X1
usb 3.1 is pci-e 2.0 X2 (may need x4)
sound chips are pci-e 2.0 X1
other Io and sata need bandwidth as well.

Re:intel's DMI bus is still to slow (1)

GigaplexNZ (1233886) | about 4 months ago | (#47015459)

Put a single Thunderbolt controller in there, and your 20Gbit allocation is already spent.

Re:intel's DMI bus is still to slow (0)

Anonymous Coward | about 4 months ago | (#47015503)

Even in theory it's not 20Gb/s, it's 16. (4 5GT/s lanes with 8b/10b encoding).
In practice, DMI 2.0 on sandy/ivy/haswell saturates at about 1.68GB/s.

Re:Obsolete tech. Pass. (0)

Anonymous Coward | about 4 months ago | (#47014409)

Not only that, but things like TRIM and GC **DO NOT WORK** through a RAID controller, especially a FakeRAID controller like the SIL that comes on these cards - with NO cache RAM.

Re:Obsolete tech. Pass. (1)

fisted (2295862) | about 4 months ago | (#47015027)

SATA6bps is the bottleneck

in fact 6bps seem kind of slow-ish

Re:Obsolete tech. Pass. (1)

jones_supa (887896) | about 4 months ago | (#47015827)

The original IDE drive interface was nothing more than an extension of the ISA bus, the primary system device interconnect, over a 40 pin ribbon cable.

Huh? No, there certainly was a storage controller chip between the ISA bus and the HDD. The term Integrated Drive Electronics refers to the fact that the drive controller is integrated into the drive, thus abstracting away the need to control the R/W head by host software.

Re:Obsolete tech. Pass. (0)

Anonymous Coward | about 4 months ago | (#47015859)

> Just this week Intel released the 9 series chipsets that support both SATAExpress and the M.2 SSD format. Bot provide 4 lanes of native PCI Express.

That is misleading. That chipset provides only 2 lanes, and only PCIe 2.0. Which means that it's already at its release too slow for some of the faster drives on the market. I.e. this chipset is an obsolete-at-release product.

And Timothy Wins the Slashdot Office Pool... (1)

Nova Express (100383) | about 4 months ago | (#47012735)

...for who can cram the most acronyms into a single headline!

Disable ads isn't working (2)

Russ1642 (1087959) | about 4 months ago | (#47012879)

I checked the box but these damn slashvertisements keep showing up.

OCZ? (0)

Anonymous Coward | about 4 months ago | (#47012881)

Not a chance in hell, even if they use Toshiba ram chips it's still the same shocking design and the same shocking lack of service.

After catastrophic data-loss (had a backup of most of it, but still takes time to rebuild) after just 160 hours of drive use these drives are worse then the Seagate 20MB drives from the middle 1980s.

Maybe, like Seagate, in 30 years OCZ will be reliable, but until then data is too precious to risk.

5 year old hardware in benchmark comparison (1)

alphad0g (1172971) | about 4 months ago | (#47012895)

No one can argue that Fusion-io started the PCIe SSD market - many laughed and now many are competing. I won't say they are the best for the price, but Tom's hardware is misleading people when they compare OCZ against 5 year old hardware. The 160gb iodrive is the original product that FIO launched with. Still a good product, but that is like comparing the top of the line pentium with today's CPUs.

Ha, OCZ (0)

Anonymous Coward | about 4 months ago | (#47013279)

I loved when I used to point out OCZ had high failure rates years ago when they were the most popular SSD and people use to just mock me as a hater. Meanwhile their SSD's all failed and my Intel and Samsung ones kept going just fine.

Why does MojoKid only submit links to HotHardware? (5, Interesting)

Aryeh Goretsky (129230) | about 4 months ago | (#47013627)

Hello,

Why does MojoKid [slashdot.org] only submit articles which link to HotHardware reviews? Is HotHardware a Dice.Com site? Is MojoKid a Dice.Com employee?

A disclaimer would be nice about paid editorial content or when linking to sister sites in the Dice Holdings portfolio, etc.

Regards,

Aryeh Goretsky

Re:Why does MojoKid only submit links to HotHardwa (0)

Anonymous Coward | about 4 months ago | (#47015847)

It seems that you have caught an actual shill in action.

Re:Why does MojoKid only submit links to HotHardwa (2)

dave420 (699308) | about 4 months ago | (#47016057)

There was a Wikipedia user called MojoKid HH [wikipedia.org] who created the HotHardware wiki article. Coincidence? Who are we kidding. HotHardware is not affiliated with Dice - it seems we just have a single person trying to drum up some traffic for their website. MojoKid's wiki user page contains comments he's made which refer to HotHardware as "us", meaning he's at least affiliated with the site. I know Slashdot has gone down the pan recently, but this is not Dice's doing.

no support, no buy, stupid slashvertisments (1)

Gravis Zero (934156) | about 4 months ago | (#47014765)

"Linux Support Planned"

color me shocked when it turns out to be closed source.

OCZ drives are just evil. (1)

ZorinLynx (31751) | about 4 months ago | (#47014933)

We had an OCZ drive fail at work and kill the iMac it was installed in.

Yes, KILL. The machine would no longer power up at all.

At that moment we didn't know it was drive's fault, so we moved the drive to a different iMac. (These are older iMacs, out of warranty.)

*POP* a second dead iMac.

I will NEVER buy an OCZ product as long as I live. I don't know how the heck the drive killed the machine, and I have no easy way to find out. Maybe I'll sacrifice an ancient PC to see if the drive kills it as well.

Performance (0)

Anonymous Coward | about 4 months ago | (#47015765)

Did anyone else notice that the advertised 'up to' 140,000 IOPS at 4k only equals 560MB/s performance? What gives?

Load More Comments
Slashdot Login

Need an Account?

Forgot your password?