More Good Things

As you may have heard, Micron recently announced its RealSSD™ family of Solid State Drives (SSDs). Beyond the endurance and reliability bonus that a solid state drive offers over ye olde spinning platters, these drives also offer substantial energy savings vs. standard HDD products.

For example, in a notebook, the SSD will consume roughly half the energy of a HDD.  Read: battery life

Now–when compared to enterprise-class HDDs, the SSD consumes about a 1/10th of the energy. Seriously. Read: Savings implemented across an entire enterprise

Why the difference? In the enterprise use, HDD products are higher performance than standard HDDs, but this performance comes at the expense of greatly higher energy consumption.  Not so for SSDs.

So, here’s another solution that adds to the larger energy efficient eco-system. Look for it in hardware coming to you soon.

Justin Sykes, SSD Product Manager, Micron Technology, Inc.

Sustainable IT

We recently spent some time talking with Ted Sampson over InfoWorld
about the role of Memory in Server power consumption. He writes their Sustainable
IT Blog … Check
it out.

Kevin Kilbuck, Senior Marketing Manager, Micron Technology, Inc.

Turns out, you don’t know what you don’t know. Really.

So we just recently we had a meeting with some of the guys that run the data center here at Micron. One of the topics of discussion was the role of memory as a power consumer in those big-iron systems … basically we wanted to get their take on memory and also get a grounded opinion from guys in the trenches on all the buzz around green(ish) data centers.

Needless to say, it was an enlightening conversation. For me, the big take-away was that while these operations guys were looking at many different ways to reduce power, they hadn’t really dug into the consumption specs around memory yet. Their assumption (and probably the assumptions of most people for that matter) was that the power requirement for memory "is what it is" …

So, one of our smart-guy engineering types heads for the whiteboard (frankly he’ll do that any chance he gets) and starts cracking off equations that calculate the benefits of dropping the voltage from 1.8V to 1.5V, and the roll of reduced chip count (RCC) modules. It really opened everyone’s eyes.

So–we tossed all of this into a really simple calculator that will take your total data center capacity and your local electricity rate and do the fancy white-board stuff for you … Pretty cool. And funny really, brings home that old cliché about change beginning at home.

Second-Sourcing Makes the World Go ‘Round

Technology takes time. Moore’s Law tells us that. But just how much time it takes for a technology to emerge that is smaller, faster, and better than its predecessor can be dependent on variables that Moore himself could not predict. In the semiconductor industry specifically, standards adoption, followed by second-sourcing, are seen to be two momentous forces that push new technology forward—pushing it to the point where it can dramatically shape its world and the products built around it. But it doesn’t always happen in that order. And who would have thought that energy efficiency in these tiny chips that draw only milliwatts of power would be as significant as we have shown?

In April 2007, Micron introduced its Aspen Memory® family of energy-efficient products, featuring the industry’s first low-voltage (LV) DDR2 DRAM in reduced chip count (RCC) memory modules, lowering DDR2 voltage to an industry-leading 1.5 volts.

Recently, there was a second-source announcement for 1.5V DDR2 and 1.5V FBDIMMs. I’m not able to name the company here, but it’s not just a DIMM manufacturer using Micron DRAM, but a company that turns bare silicon wafers into DRAM chips. There are now two companies using a sub-80nm process to get to 1.5V. And all this before energy-efficient standardization has found top billing on JEDEC’s agenda.

We are not trying to be critical of JEDEC. After all, standards are necessary to ensure interoperability. The standard for 1.5V will not define how much power the device can consume, but rather that all 1.5V devices work together. Before a standard makes it on the agenda, there has to be a need—and the market needs to define that need (lower power consumption). Data center operators ask server manufacturers for more efficient systems, and from there, the requirement works its way upstream to the memory industry.

So, we have the typical "what came first" paradox. Instead of trying to predict how this will play out, let’s consider some of the common memory products that were widely adopted before they were standardized by committee:

• PC100: PC100 was unilaterally defined by Intel® in 1998. It was close to the then-defined committee standard, but Intel said it was leaving too much performance on the table, so they tightened it up and squeezed out the technology laggards who weren’t able to make the product. Those who couldn’t yield to PC100 were stuck making PC66 a little bit longer than they expected.
• DDR-400 (a.k.a. PC3200): Became an industry standard outside of committee. In fact, the standards group originally thought that 400-speed memory was supposed to be only DDR2. Sure, there was PC2-3200, but it pretty quickly moved up to 533 and 667. In fact, PC3200 needed to run at a boosted voltage to get up to speed. PC2700 runs at 2.5V while PC3200 needs 2.6V.
• DDR2-1067: This time JEDEC came up with two versions of 800 MHz memory-DDR2-800 and DDR3-800-probably hoping this overlap would stop the technology leaders from offering a disruptive solution. But, it’s hard to hold back innovation, so now Micron is shipping DDR2-1067.
• 1.5V DDR2: Of course I had to add this to the list. It may be a little premature for now, but someday soon this, too, will be shipping FBDIMMs and components in volume.

Two companies at 1.5V … I guess I could wrap this up with a this-is-just-the-tip-of-the-iceberg analogy, but if you’ve been around the industry for a while, you know it already. Second sources for innovations make the world go ’round—and the committees will simply add to the momentum.

Michael Sporer, Regional Sales Manager, Micron Technology, Inc.

Finding the Energy Efficiency Balance

The whole point of the focus we’re seeing on data center efficiency is because data centers consume so much energy and keep on growing. One of the drivers could be an idea that’s been around for awhile: the idea that you centralize all your computing resources and access them through a thin client or browser. It’s an idea as old as computing itself.

But aside from the focus on servers, the storage equipment in the data center is not to be overlooked either—all those spinning disks could use some help from solid state based storage.

And while we’re at it, let’s step up a level higher still. The real issue isn’t about just reducing energy consumption in one type of hardware or in one segment; we need to understand the consumption across applications and deliver an array of solutions that collectively has the highest efficiency possible over the entire product lifecycle.

So let’s keep our eyes open for opportunities and balance. If we are too aggressive, it may end up costing more ($ or kWh) than is being saved. The flip side, of course, is that we wouldn’t be aggressive enough and it wouldn’t help as much as it could. We need to find just the right balance.

Michael Sporer, Regional Sales Manager, Micron Technology, Inc.

No, This is Not My Car

Here I am blogging about energy efficiency and yet I don’t drive a hybrid car. In fact, I don’t even drive a high MPG car. Why not? Short answer: My current car has not outlived its useful service life. Yet. Stay with me for a second: a car is built, and over its lifetime is driven (most likely by more than one owner) until it is no longer serviceable. Then, it’s retired and hopefully recycled.

So, what is the energy footprint of a car? Well, it’s how much energy it takes to make it, how much energy it consumes to use it, and (in the end) how much energy it takes to recycle it. How long I own the car doesn’t really play into it at all. Although how much I drive it does. So, as my current car inevitably fades into its autumn years, what’s the next car I’ll buy? I don’t know yet. But here are the simple criteria:

• It has to meet my needs: to match the level of performance I am accustomed to
• It should be as efficient as is reasonable

This deceptively simple principle is a gauge by which we can measure anything, really. For example, right now data centers are struggling to keep a balance between delivering higher performance services and the off-the-chart energy/cooling consumption that goes along with them—and while they may not be hitting the apex of their hardware lifecycle just yet, we can begin making the design changes now that will move us into the next generation of efficiency while still meeting the growing needs of the users—this can be done.

So, will the market make a car available to me that is more efficient than the one I’m driving now? I hope so. Should the market provide an incentive as it did in my light bulb example? Hmm, to date there have been plenty of incentives offered towards the adoption of energy efficient vehicles. Do they need to continue? I don’t know, I’m not an economist. But, when the time comes in my current car’s lifecycle to transition to a more energy-efficient device, I will—and I intend to drive the new one for a long time. It’s as simple as 1, 2—right?

Michael Sporer, Regional Sales Manager, Micron Technology, Inc.

Power-Hungry Data Centers Demand Memory Innovations

It might have been at one time that a data center was a collection of servers inside a corporation; servers that do things like your e-mail and the like. But today we’ve got data centers all over the world, and they are doing things like running the Web. I was really surprised when digging into this a little bit and discovered that worldwide, data centers consume over a 100 billion kilowatt hours annually. A hundred billion kilowatt hours; it’s almost unfathomable.

In this trend of more and more rich data being moved through these server centers, it’s just all going to add up to consume more and more power. If you look at the overall piece, 100-plus billion kilowatt hours, and if you just look at the memory piece of that and think about what we can save, we’re still talking about some pretty significant numbers here. Roughly five to six billion kilowatt hours of memory power consumption could be saved just by adopting the new Micron parts the 1Gb-based, DDR2 reduced chip count (RCC) modules and 1.5V DDR2 FBDIMMs.

Dean Klein, VP of Market Development, Micron Technology, Inc.

Power Demands of Data Centers Require Memory Innovations

The U.S. EPA ENERGY STAR program is conducting a study to assess opportunities for energy efficiency improvements to computer servers and data centers. This is in response to Public Law 109-431, which was passed and signed into law December 20, 2006. This legislation requires an investigation down to the microchip level. As a manufacturer of semiconductor memory products used in server systems, Micron intends to proactively address these opportunities.

I authored a white paper called “The Power Demands of Data Centers Require Memory Innovations,” which talks about DRAM in computing applications. You can download my white paper from Micron’s Energy Efficient Memory Web site.

Michael Sporer, Regional Sales Manager, Micron Technology, Inc.

Micron Joins the Green Grid

The goal of the Green Grid consortium is to reduce power consumption at computing data centers worldwide. It’s a membership made for us. We have a major silicon component in servers. And we have ways of designing those memory modules so they will consume less power. Our new Aspen Memory® modules are prime examples. The energy savings they can provide is pretty amazing—especially on a worldwide scale. And it’s right in line with the Green Grid’s mission.

Dean Klein, VP of Market Development, Micron Technology, Inc.

What’s the Next Step to Power Savings?

Last week I got a letter from my local utility company, offering a great deal on compact fluorescent light (CFL) bulbs. The letter pointed out that if everyone in my town changed ONE incandescent bulb to a CFL, we could save enough energy to power ~1/10 of the homes in the town for a whole year.

I decided to install CFLs because it made economic sense to me. The utility company offered me an incentive to do it. Does that make me any less of an environmentalist? I’d like to think not.

Needless to say, the postcard went into the recycle bin because I was already doing my small part. But beyond my four walls and the town where I live, big server farms and data centers are consuming billions of watts of power. With the right tools—like power-saving memory modules—and the right incentives—like bottom dollar savings—then I think we’ll start seeing real results.

Michael Sporer, Regional Sales Manager, Micron Technology, Inc.