Micron Demos Blistering 1GB/s Solid State Drive

Micron Demos Blistering 1GB/s Solid State Drive

When it comes to memory technologies, there are a few major names in the industry that are essentially the primary sources of the chip-level technologies that drive the market.  When it comes to DRAM and Flash memory, Samsung, Intel and of course Micron, are a few of the big names.  Lately, though DRAM technologies have taken a sleepy development curve, Flash memory for SATA-based Solid State Disk technologies, has been on fire. Recently we showed you what Intel's new X25 series of SSDs could do and without question, they're some of the fastest SATA Solid State Disks on the market.  In addition, there are various incarnations of Samsung's SSD product on the market and they offer potent storage solutions as well.  Though with any current traditional SSD technology, random write performance seems to be the Achille's heel versus traditional spinning media. 

Micron has been relatively quiet with respect to SSD technologies, however, that may change dramatically in the not so distant future.  Recently Micron demonstrated a dual processor, eight-core Xeon workstation with a pair of PCI Express-based SSDs installed.  Since the drives are directly attached to a PCI Express serial link, they were not limited to the 300MB/s bandwidth cap of the SATA interface.  What's more interesting is that the drives also incorporate data management algorithms that mitigate random write latency and increase performance in general. 

Micron's Joe Jeddeloh demonstrated in the following YouTube video, a single SSD that achieves 800MB/s and 150K IOPS throughput, as well as a dual card configuration capable of 1GB/s and over 200K IOPS.  This was achieved over an 8X PCI Express link in a full length card.  The technology is impressive to be sure but video of the IOMeter test is a little on the fuzzy side...



Jeddeloh notes that the dual SSD card will be "coming to you soon" and we can't help but wonder about the cost.  Regardless, with the kind of performance that is demonstrated here, the technology is obviously a significant break-out from the SATA paradigm that currently most all SSD manufacturers are working in, save perhaps for the folks at FusionIO who have a similar PCIe-based technology. 

Another blogger on Micron's site goes on to say, "...we have plans to deliver this product to selected customers in 2009, with wide availability in 2010. Unfortunately this specific product is not targeted at your laptop (unless you have a very big laptop!) but this is a good indication on where this technology can take us in the future. It is within the realm of possibility to deliver this type of performance in a laptop if the architecture would support it. Currently available buses in laptops would not keep up. In fact, desktop buses don’t keep up, either."

It's exciting to think where storage technology is headed and without doubt, the end of the line for rotational media is clearly on the horizon.

0
+ -

Gotta love the fiendish grin on JJs face the whole time. lol

0
+ -

See this is what I'm looking forward to. Why stick to the harddrive format if you don't have to. Can't wait till this market matures a bit.

0
+ -

They may need to create a new mobo standard(way longer board)with nothing but pcie slots... like a dozen of em at least if not 20 or more. Gonna need a taller tower too... like 4-5ft tall maybe even 6. Indifferent

How many extra points do I get for blatent sarcasm?

0
+ -

recoveringknowitall:

 

How many extra points do I get for blatent sarcasm?

15

Not even just PCIe, but there is no reason to stick the the Hard drive format. It's kinda a waste of space if you think about it. You could even plug them in like sticks of RAM.

 

0
+ -

Rotational media as you call it will not be going anywhere, despite what some of you fools think. We still have tape drives even though we have new fangled optical media that beats tapes hands down in speed but not in capacity. While what micron has done is very nice what no one is talking about is the fact that SSD's have a FINITE number of write cycles before that bit becomes useless. Sure they quote huge MTBF numbers, but hell i doubt that anyone here doesnt know of someone or had their own flash drive die on them. It is the same tech, and while flash drives are cool and last a long while, eventually you will wear them out, and relatively quickly to. Windows swap file hmmm thats alot of reads and writes, and with no SSD's out long enough to see if they even last a year or two under "normal" use.

Disk drives will still be around for decades to come due to the fact that they are a mature platform and have been operating for years with server drives able to run for 10+ years under heavy loads of read and write access. SSD's have a very long road ahead of them if they wish to take over as a primary storage medium in computers, that is reliable and will not wear out in a couple of years of use.

0
+ -

Drago, I don't completely agree with that view but do agree that standard HD technology will be around for a very long time. However, SSDs are going to slowly eat into their market share, as the technology ramps up, becomes more cost-effective and reliable. It's just a matter of time.

0
+ -

And if you don't think standard HDs don't have a finite number of write cycles, you're sadly mistaken. :) In fact, once the wear-level alogorithms are improved and the memory is developed with better endurance, mechanical media will become far less reliable due to its susceptibility to shock and vibe. Again, it's just a matter of time.

0
+ -

Drago:
despite what some of you fools think
There are no fools here at HH IMHO. :)

0
+ -

At some point or another, "everybody plays the fool", "no exceptions to the rule".

Now that I've got that out of my system, I have to say that I firmly believe over the course of the next few years, SSDs will absolutely replace traditional HD's in consumer level computers. Maybe not for storage needs, but certainly for boot drives.

0
+ -

It's 1 thing to play the fool and another to be a fool 24/7... and trust me I've known a few.

Now that I've got that outta my system, I can safely say that everyone has made valid points.

The bottom line = We're all crazy about technology and hardware and want nothing more than to see it advance and evolve.

BTW: Gotta agree with Dave: Drago rox for inciting some lively discussion!

(I'm fairly certain that was his goal rather than trying to be condescending)

0
+ -

I like Drago a lot and I dont think he was being condescending at all. This is a good conversation in my opinion. I think high end systems will be running SSD boot drives very soon. It will be years before it hits the mainstream. Laptops might get them sooner. May not be the fastest of SSDs, but they have more advantages for lappys than speed.

0
+ -

Good performance, but SSDs will probably be for high end PC's until they actually become a viable replacement for HDDs.

0
+ -

Drago -  How long did it take for punch tapes to be replaced by Magnetic tapes? For Magnetic Tapes to be replaced by the 24 in. floppy?  The 24" floppy by the 8" floppy? The 8" floppy but the 5 1/4" floppy? The 5 1/4" Floppy by the 3.5" Floppy? How long did it take for the DVD to hit the market? And from the DVD to the blu-ray? 

Do you see where I am going with this?  Every generation of media has been replaced a few years quicker than the previous one.

What you don't seem to understand is the MTBF numbers and how they are developed.  What they do is they put it under the absolute maximum read/write load that it can handle.  Then they let it run until it dies. And they don't do this with just 1 or 2.  They do it with a couple HUNDRED. When each one dies, they record the total number of read/writes and the time that it took for the SSD to fail. Then they compare the total number of read/writes they got to the average number of read/writes that windows performs per hour and they can establish their baseline and be very confident about it.  Its very similar to the testing that auto manufacturers do for testing paint jobs to see how long they will last in certain conditions.

The HDD, as we know it, has been around for over 3 decades.  If you think its another 2 decades before its gone, then there is a serious problem with the computer industry because the advancement will be lagging behind.  Either that or the world as we know it came to an end..

0
+ -

Like drago said tape drives are still around, but how many of us are running them? I don't expect HDs to go away any time soon, but the consumer market moves much faster that the server market.

0
+ -

All I have to say is Drago rox for inciting some lively discussion! :)

0
+ -

Another frequently overlooked solution is ramdisk technology.

Yes, all RAM is volatile, but solutions to that problem are

now readily available e.g. RamDisk Plus from www.superspeed.com

supports an option to save and restore ramdisk contents

between shutdown and startup.

Version 9 also supports creation of ramdisks in unmanaged

Windows memory e.g. RAM addresses above 4GB with XP x32.

Intel CPUs have supported 36-bit hardware addressing

for a very long time now!

A second consideration is the efficiency of RAID controllers:

scaling RAID 0 arrays from 2 to 4 to 8 conventional hard drives

will almost always demonstrate a ceiling imposed by the

RAID controller, NOT by the media. And, the same is true

even if multiple SSDs are assembled in a RAID 0

(1 Super Talent 16GB SSD @ 130MB/second; 2 @ 150MB/second RAID 0,

measured just last week).

A third consideration is chipset efficiency:

some "x8" and "x16" slots only get x4 PCI-E lanes,

and that can be a very big surprise for designers

who haven't taken that factor into consideration.

Now, having said all of the above, high-grade RAM

typically comes with a lifetime warranty and

has no write-cycle limitations.

So, consider this: DDR2 is now about one-half the cost of DDR3,

and the P45 chipset supports 16GB of DDR2 e.g. ASUS P5Q Deluxe.

At very competitive prices an XP 32-bit or 64-bit system can

support a relatively high-speed ramdisk of 12-14 Gigabytes

that is effectively NON-volatile (16GB - 2 or 4GB for the OS etc.)

Yes, I realize that many IT users want the whole enchilada:

multiple terabytes ALL accessible at one gigabyte+ per second.

But, such expectations are not only unrealistic but also

unnecessary to achieve super high-speed I/O -- particularly

with a little ol' fashioned "data management".

One last item: MetaRAM now builds R-DIMMs at 16GB per stick!

Consider the possibilities e.g. if that technology were implemented

in SO-DIMMs, or in DDR3 DIMMs that work with Core i7 w/ 6 slots.

6 slots @ 16GB = 96GB per ATX motherboard!

What am I missing here?

Sincerely yours,

/s/ Paul Andrew Mitchell, Inventor and

Systems Development Consultant

All Rights Reserved without Prejudice

0
+ -

Drago there is a finite point your missing here on SSD technology. That would be the new IBM Blade Servers. The problem with Standard harddrives being there energy usage cap. As has been illustrated in many areas of data in a data base/warehouse or large media manipulation operation (film studios, video game development, video manipulation storage etc. etc.) there problem is not length of usage time or size; it is energy! Because a SSD blade server uses 5% of the energy used by a standard server it and reiterations by other server manufacturers is going to become the standard at least on that end. The reason for this is energy caps and platform upgrades. Many giant operations of the type I mentioned earlier are hitting a brick wall. That brick wall when the energy provider says I can't give you any more electrical energy in your current facility. At that point slowly but surely these organizations are changing over because a 95% energy reduction means they can almost fully double there standard size. The effect this has is a trickle down effect in prices. As this multi pica byte server market emerges and is implemented the commonality of ssd's changes in numbers, therefore the prices drop and the size goes up as well as all the other implications (speed, reliability, size). I give it 5 years and think that's stretching really

Login or Register to Comment
Post a Comment
Username:   Password: