Imation partners with Mtron, tosses its own hat into the SSD ring

SSD is really going mainstream this year, and while Mtron has been showing up with the goods -- and powering our favorite RAID array ever -- it looks like it took a brand like Imation to take it to the next level. The two companies are teaming up for two new SSD product lines: Imation SSD MOBI 3000 for laptop and mobile use, and Imation SSD PRO 7000 for enterprise purposes. The MOBI 3000 sports 100MBps read times and 80MBps write, with 0.1ms random access times, which supposedly makes it the fastest SSD product on the consumer market -- the drive purportedly cuts boot times in half. The SSD PRO 7000 does 120MBps read and 90MBps write, and has a mean time between failure greater than one million hours. Price and availability will be announced in the "coming months." It can't come soon enough, believe you us.

[Loonie @ Jan 4th 2008 1:46PM]

I do so hope "mainstream" means "affordable".

[Faisal @ Jan 4th 2008 1:49PM]

Out of curiosity, how do you measure a mean time between failures that's a little less than 115 years?

Silly questions aside, this is pretty exciting. I'll be able to finally...well, no idea yet, but I'm sure it'll be obvious once the potential's there.

[Mike @ Jan 4th 2008 11:16PM]

It's all theoretical based on lab stress tests and some kind of use case that is often poorly defined or unrealistic.

Sadly, a lot of SSD manufacturers determine MTBF numbers with the test criteria simply being "powered on" - with no mention of read/write cycles.

[Andir3.0 @ Jan 4th 2008 2:29PM]

How do you measure the age of dinosaur bones? galaxies?

[Faisal @ Jan 4th 2008 3:14PM]

Agreed, Mike; I think that was what I was trying to convey, since getting a figure like an average of 115 years for continuous, fault-free operation indicates that there might be something wrong with the model.

Dinosaur bones and galaxies have been around for a while...ascertaining their age is kind of the opposite of ascertaining how long something will last in the future, I'd assume.

[Andir3.0 @ Jan 4th 2008 3:42PM]

It all amounts to estimates based on some form of measurement, be it carbon dating or a sample "abuse" test of read/write tests.

[Maverickfiveo @ Jan 4th 2008 5:03PM]

I agree. How does someone come up with an MTBF approximately 100x times longer than the device has been in development? I don't mean this to be a snide comment on Imation or anyone else: I'm genuinely stumped. How are these things calculated? Anyone know?

[reliability @ Jan 4th 2008 5:17PM]

MTBF should not be confused with life expectancy.

MTBF is a statistical value which indicates reliability. Lets take an example: If we have 100 000 unit population of identical devices with MTBF of 1M hour. It simply means that one device gets broken every ten hour starting from hour one.

[Ed @ Jan 5th 2008 5:58AM]

@reliability

That is a very good and interesting point. However, it is very easy and understandable to confuse it. I myself, did not make the distinction until I read your post.

Additionally, the layperson only cares about how long the device will remain usable, not how long the parts would last on their own. I assume that is how you are defining life expectancy, since you are making a clear distinction between reliability of its function and the life expectancy of the object.

Sure, MTBF is a statistic. You do a good job of explaining that their is difference and a great job of explaining the statistic itself, but I would argue that there is effectively, no difference to the average person when talking about MTBF.

When I chuck the hard drive into the desert for not working anymore, I am not particularly interested in how long it will take to degrade.

P.S - I'm kidding about chucking it into the desert. I use old hard drives to make furniture out of them. More environmentally friendly :)

[reliability @ Jan 5th 2008 7:09AM]

Greater MTBF usually means greater life expectancy but they do not equal for the very reason why the device would not operate over 100 years.

Most MTBF calculations take assumption that during normal operational life (useful life) we are in the middle of the bathtub curve where constant failure rate exists, i.e. It is equally probably that the device gets broken at any given time. It is impossible to predict when one particular device gets broken but we can pretty accurately predict behavior of big population. This basically applies to any statistical mathematics.

http://en.wikipedia.org/wiki/Bathtub_curve

[Nubaeus @ Jan 4th 2008 2:25PM]

"It can't come soon enough, believe you us."


I only have one thing to say:


?

[Phil Perman @ Jan 4th 2008 3:17PM]

Its an adaptation of "believe you me", a fairly common expression

[Flashpoint @ Jan 4th 2008 2:35PM]

A million hours = 41,666 days

How could they possibly ensure 1,000,000 hours between failures?

[Ricardo @ Jan 4th 2008 2:43PM]

Looks like 2008 will finally be SSD year!

[rv @ Jan 4th 2008 3:19PM]

The only thing better than a ssd... is more ssd's.

[axelkung @ Jan 4th 2008 5:02PM]

http://www.bitmicro.com/products_edisk_altima_25_sata.php

it says,"2.5-inch: 8 to 416 GB*....Up to 64 GB at 8.5 mm height" (416/64*8.5/2.54 is about 2.17-inch)

So, 2.5-inch also means its thickness??

Hmm......give me a "brick".

[Ricardo @ Jan 4th 2008 6:04PM]

"Up to 64 GB at 8.5 mm height" doesn't mean it's 8.5 mm for each 64 GB. It just means bigger drives are taller, but doesn't specify how much.

[Jeff Lewis @ Jan 4th 2008 5:08PM]

Wow.. this will be fun once

a) They start actually shipping,
b) We finally get a price.

You'll notice that so far, no one is selling these things retail and no one has a price, other than the Dell 32GB SSD that goes for around $450. Samsung only sells to OEMs as far as I can tell.

When I can go into Best Buy (or whatever electronics chopshop you like) and buy one over the counter, I'll be more interested.

Until then, this is basically vapourware with really good photos.

[Ed @ Jan 5th 2008 5:15AM]

OMG.... SSD is such bullshit. I have never seen anything with so much propaganda that is only hyping a crappy product. Or should I say, crappy with respect to how good it was supposed to be. In some ways an SSD is WORSE then a spinning hard drive.

I, like many others, had a 6ft boner when I heard about these devices and was nearly ready to kill to get my geeky hands on them.

What a TOTAL letdown.

The NUMBERS are NOT REAL. They are giving you a super rosy picture about how well they could operate under 100% ideal conditions. Solid State sounds good, since they always talk about a hard drive falling. That's like someone slipping on a banana peel. We have all heard about it, seen cartoons with it happening, but have you ever heard of someone actually doing it? I have never actually heard about anybody losing data from a hard drive because it fell on the ground. Not once. Seriously. I know it may have happened to some people, but I don't know them. Does not sound to me like a reason alone to move to Solid State, more like fear mongering advertising tactics. I have a cream which will keep your pecker from getting SMALLER too. Just send 5 dollars to.... :)

MTBF?

Based on ideal conditions, like performing a couple of read/write operations every day. I bet they are only using less then 10% of real world conditions for estimating their MTBF. They are not clear about what percentage of the memory is unusable either, since it is kept for bad blocks, or to replace bad blocks that inevitably appear due to the inherent write fatigue in flash memory. We only use flash memory occasionally to transfer large files around. Try using it as a professional under heavy use. They wear out. Using it as a main volume for an OS would take that to a use that is well above and beyond the normal use for one of these things. So the MTBF would have to take into account the inevitable wear on the memory and lost memory blocks. That means the hard drive would SHRINK. Try including that in the MTBF. Maybe it's up for interpretation to say a hard drive has failed when 20% of its space has disappeared.

I do know this. RAID does not like hard drives who change their sizes. Not a big concern over the course of a couple months, but 3 years? 4 years? 5 years? I dunno. We will have to wait and see how they really perform over that time period. At least MTBF on spinning hard drives has been evaluated for more then 10 years.

We also don't really know what measures they are taking to address this yet. It's all 1st gen anyways. I just know that the real world results of these things NOW is not as good as it was supposed to be.

Read Speeds? Latency or Throughput?

Any layperson can understand that a spinning hard drive is like a 4 lane freeway doing 60 mph and memory is like the Millennium Falcon doing the Kessel Run in less then 4 parsecs. Yawn. Of COURSE read speeds are going to be faster. The latencies are just sick. They could push more then 100 MBS (throughput), but that is not a limitation of SSD, but rather a limitation of the interface, SATA, PATA, etc.

They may state 80MBS as a write speed, but lets see an actual 3rd party independent benchmark test doing sustained writes. 4K, 16K, 64K, 256K, etc. Let's see how it really operates under light, normal, and heavy conditions.

I have seen actual tests that show that SSD's can be up to 8X SLOWER then a spinning hard drive with random writes less then 50K. Google for it.

This is being over hyped to the consumer, specifically gamers, that it is a great solution for sick performance levels. The read speeds to do help load those levels. That's true. However, OS load times are not that much better according to people I have actually talked too that own one. It's just not worth that much money to get a fraction of the space of a spinning hard drive.

Corporations? Forget about it. I can run firebird databases with 250 million rows in them on a 15K SCSI SAS ARRAY that will STILL outperform an SSD RAID array. Why? Reading the index on a database like that will happen a LOT quicker on an SSD. However, that will only affect reads on a database. Inserts and Updates will actually happen slower. A database of any considerably size under normal use also generates a lot of IO's. Bad for flash. The constraints and triggers on your inserts and updates may happen faster, since the reads improve, but that is all taken away with the slower write performance. If an SSD is really 20X faster on the reads and 8X slower on the writes, do the math. 1 second for reading and 1 second for writing. 2 seconds total normally. SSD will make it 1/20th of a second for reading and 8 seconds for writing. 8.05 seconds total. Now that may be overly simplistic, but prove me wrong with real tests :)

I just wish some of these websites would start posting articles with links to real world tests and stop hyping the bullshit.

Let's see Engadget post an article referencing the REAL performance levels of SSD's.

[Mr. Deeds @ Feb 23rd 2008 1:47AM]

Type "MTron" into the search area at youtube.com.... I don't think you're gonna find a "Spinning" drive boot XP in 6 seconds from BIOS post. There are plenty of videos for you to check out for proof of SSD performance. Talking about load rates changing performance??? We're talking about a product that doesn't move, and it's called "Solid State"... It's just "there". There's no burst rate, and no changes. It's a consistant rythem that never changes.

[Ed @ Feb 23rd 2008 4:29PM]

@ Mr. Deeds.

Your just like all the rest. "Proof of SSD Performance"

Okay. Would you mind breaking that down into WRITE SPEEDS and READ SPEEDS?

The fact that is Solid State is irrelevant. It could be Tapioca Pudding State, Or Millenium Falcon Marshmallow State, or even Strawberry Shortcake Yummy Goodness State.

If we are going to look at performance, you need to look at write speeds and read speeds, among many others.

You cannot get around it. If you actually perform any tests on a 1st Gen SSD drive, you will find write speeds performing POORLY.

Period.

[porath @ Jan 8th 2008 12:33AM]

first of all, these "newly announced" drives seem to just be rebrandings of the mtron drives. is that not the case?

second of all, the mtron drives *are* available right now. here's one site you can buy them from:

http://www.ssdisk.eu/