Special Today: this report on last week’s Flash Memory Summit is written by Steve Denegri, a long time observer and analyst of the storage scene. The snarky paragraph heads in bold are mine. Thanks, Steve!
As if the hard disk drive industry didn’t have enough anxiety already. . . .
It appears as though flash memory will expand its target market from cell phones, digital cameras, and iPods to include notebook computers in 2007, coincident with the launch of Microsoft’s ‘Vista’ operating system. This was the obvious conclusion drawn by attendees of the Flash Memory Summit in San Jose last week.
It seems like a big jump for flash, particularly since we, the users of flash, more closely identify this technology with capacity points in the 1 to 8 gigabyte range. And while the capacity of flash is doubling every 6 to 9 months, it certainly doesn’t seem as if we’re ready to toss out the hard drives in our notebooks for flash, does it?
Peaceful Co-existence.
Not yet, it doesn’t, but does appear as though flash has now established for itself a proverbial “foot in the door” of the notebook PC market. As a first step in what appears to be a multi-year evolution, flash will coexist with the hard drive, and the first phase begins next year when a small reservoir of non-volatile memory is implemented for the purpose of more rapid boot of the Windows-based notebook.
The faster-performing flash component will help Microsoft address the mounting problem of progressively slower notebook boot-up as the Windows O/S has advanced. The primary culprit to date has been the reliance upon the slower-performing hard drive relative to system memory for intensive boot tasks. For example, with Windows XP, 80% of system memory is unused at startup, leaving the hard disk to do most of the work.
Windows Bloat: Get Help In A Flash
Each new Windows release entails millions of incremental lines of code that require increasingly-spacious storage capacity, even as notebooks have trended toward more compact, more power-efficient designs. This dichotomy has claimed the disk drive vendor as its victim, primarily because the spatial limitations associated with smaller notebooks has resulted in slower rotational drive speeds. The result is a price premium for notebooks relative to desktops that evidences itself in terms of portability and convenience, but not performance.
Flash Hide and Seek
So the debate in the computing industry is around where to place this flash reservoir. Two approaches have emerged. One is to embed a 256MB or greater write buffer of non-volatile memory directly on the circuit board of the hard disk drive, a program Microsoft has labeled “Piton”. This “hybrid” disk drive essentially bifurcates the storage needs of the notebook, whereby flash is used for random operations and the disk platters are the source of lengthy sequential operaitons. For example, boot and system resume functions would be handled by flash but storage files like MP3’s or video clips will still be stored on the disk drive.
Your Money Or Your Flash
The only issue with the “hybrid” disk drive is the cost of the new memory component. The $5-10 cost might simply be too burdensome for a hard drive industry that’s already plagued with razor-thin profit margins. And some suggest Hewlett-Packard wants a 512MB buffer, at a minimum, for its notebooks. This certainly won’t help the cost equation.
Then there’s the alternative approach, which is to solder the non-volatile memory right on the motherboard. Intel has such an objective with its “Robson” initiative. In its current form, Robson is a PCIe half-minicard equipped with 1GB SLC-based NAND flash, a control ASIC, and a driver. However, by the time it ships in 2007, don’t be surprised if it’s 4GB and motherboard-centric. Some suggest that Intel is presently behind on its plans to bring this motherboard implementation to market, but the good news for Intel is that as long as Vista is the targeted application, they probably have more time than they think to get it ready for launch.
Competition: How Sweet It Is. For Us
So the mere fact that we are discussing a need for flash as a storage option in notebooks must send shivers down the spines of the six providers of mobile disk drives. And now that flash has an entrée, its vendors are raising the rhetoric by prophesying that notebooks will eventually rely solely on flash for storage needs once we cross the 32 or 64GB threshold for NAND. At the current pace of flash innovation, we’re only two or three years from reaching this inflection point.
We Don’t Need No Stinking Disk Drives
Notebook vendors seem willing to test users’ interest in notebooks that are flash-only. At the Summit this week, Samsung was demonstrating the benefits of solid-state drives in Dell notebooks. This might be Dell’s way of using a common notebook ATA socket for whichever the customer chooses, be it a hard disk or a solid state option.
Petite Power Users Rejoice
What might attract users to a notebook PC equipped only with flash as a storage repository? One reason might be power usage and battery life, but there seemed to be little agreement about that at the conference. There was a panel discussion where I asked the question. One guy from a company that makes SSDs said there was a 20-30% savings in battery life with SSD vs HDD, then this guy from mSystems rebutted it was much longer than that….3x as much, which I found absurd. The next panel discussion, someone in the audience threw out the 5% figure, and no one from that panel (including a different mSystems guy and one from Intel) seemed to take issue. Samsung claims its SSD’s power consumption is 5% that of a comparable HDD.
Carry That Weight
How about the weight of the machine? We’re really only talking about two or three ounces, depending upon whether you compare against 2.5″ or 1.8″ mobile drives, so there’s hardly a noticeable difference. Although if you factor in a lighter battery it could start helping.
A Pearl Of Great Price
Might flash be cheaper? Hardly, flash is much more expensive at $13 to $15 per gigabyte. As a comparison, you can pretty much buy as much hard disk capacity that’s available for the price of 32GB flash drive.
End of discussion, some would say. With uncertain improvement in battery life or machine weight, combined with the fact that I’m going to spend more money to get less storage, this makes for an easy decision. Besides, 32GB doesn’t give me much room to continue downloading music and video at my current pace, so why bother? The end user who wants a lot of storage capacity in a smaller, more power-efficient package will likely favor the notebook equipped with the disk drive, particularly once 1.8″ hard disk drives begin shipping in higher quantities.
They Don’t Call It Flash For Nothing
One possible advantage didn’t get much play at the summit: Flash may offer significant performance advantages. The Samsung 32 GB flash drive claims 300% faster reads and 150% faster writes than 1.8″ disks. So systems boot faster, applications load faster and large documents – like big presentations – should appear faster as well.
IT Might Like This
On the other hand, the business user may favor the flash-based notebook over the hard disk alternative. How will this be the case? Ask any help desk or IT support professional how appealing a lower-capacity, disk-free notebook would be in the hands of the corporate user and you may have to mop their drool off the floor. With the rampant pace of disk drive failures and virus attacks, combined with the public relations headache that’s often times associated with notebook theft, the flash-based notebook looks like a terrific alternative. If the user could rely more upon the corporate network for storage, thereby minimizing the amount of data that’s resident on the laptop, there emerges a reduced security concern, a very valuable proposition to the corporate IT department. But is 32 or 64GB sufficient to run Microsoft Office, Adobe Photoshop, and a web browser? In most cases, it probably would be.
See Your Flash Life Before Your Eyes?
These potential customers do have one primary concern, judging by the attitudes of attendees at the Summit, and that’s the endurance parameters of flash memory. Just like those batteries in disk-based iPods could only be recharged so many times, so can flash only erase and write data a finite amount. Furthermore, many at the conference suggested that vendor-provided parameters regarding their own products’ endurance was too generous, and in some cases, it was said to be far too optimistic based upon laboratory tests. At the present time, there do not exist sufficient standards in regard to the establishment and governance of flash performance benchmarks, illustrating the still-immature nature of the market. However, the industry is rapidly moving to solve this issue and, once complete, it will gl a long way to foster the effort to replace disk with flash in notebook PCs.
Predictions
So let’s make a few predictions. First, hybrid disk drives might get the flash movement rolling within notebooks, but these designs won’t likely be around for very long because that non-volatile memory cache will quickly move to the motherboard, perhaps after only one or two product generations. Second, look for the ultra-portable end of the notebook market to favor flash over disk fairly quickly, especially with the business customer given that it simplifies IT support issues and helps sidestep the problems associated with data theft. Third, look for flash to become a lot more affordable once it begins appearing in notebooks in widespread fashion. If there’s anything in computing that can be counted on its dramatically-lower price points as technology is standardized then, eventually, commoditized. The memory market is certainly familiar with the nature of this trend, particularly over the last decade.
Steve Denegri
Independent Data Storage Analyst
Comments welcome.
StorageMojo.com’s Take:
Odd, isn’t it that the flash folks don’t seem to have a clear idea about the power benefits of flash? I’ve done the math so you can see what benefit is – or at least see how it is done – and do a better job of it yourself. If you do, please post your results in the comments here.
Also, I think lots of road warriors will be fine with 32GB, even though Vista weighs in at 14GB. So does Apple’s Mac OS X, but after tossing out several GB of printer drivers, foreign language fonts, trial versions and other gunk, it almost gets svelte. 10 GB for the de-gunked OS, 10 GB worth of applications and 10 GB for documents, means that a 32 GB SSD can easily handle a road warrior’s requirements. Sure, add a lot of music, photos and some movies and you are way beyond that, yet if you are mixing that much pleasure with business you aren’t, in my experience, working with corporate assets. Buy an iPod for the personal stuff.
Finally, it does appear that flash’s $/GB is dropping faster than that of hard drives. As I concluded in Seagate’s CEO Dreams of Big Disk Market & High Stock Price
Flash prices today are only 2.5% of 2001’s prices. Since flash prices are driven by semiconductor technology, primarily larger wafers and smaller features, that trend will continue. So today’s $20/GB flash drive will be a $0.50/GB drive in early 2011.
Assuming Mr. Watkins is correct about hard drives, late 2010 will look like this:
- A 200GB 2.5″ drive will about $40 (applying his reasoning about 1″ drives)
- A 100GB flash drive will be about $50
- Ergo, flash will own everything under 100GB
This will be very good for us consumers – and almost as good for the one or two firms that win the fight for market share.
Comments always welcome
Why isn’t this happening already? Am I the only one who remembers 340MB and 810MB drives in notebooks in the mid ’90s?
I’m pretty sure that flash will be used more and more. Though the endurance of flash memory over the long term is a concern. File systems would have to be modified so that they try to use flash memory uniformily, rather than writing to the same areas over and over. I believe Intel/MS have something in the works to do this. I think? M-Systems have such technology as well (see http://www.m-systems.com/site/en-US/Technologies/Technology/TrueFFS_Technology.htm)
So, some work to do still.
The reason it’s been slow to develop, IMO, is because we needed a big vendor to find an application to “push” flash to customers, I’d guess you’d say. In addition to faster boot, do some web digging for Microsoft’s “Superfetch” technology embedded in Vista, which takes advantage of flash’s high performance. It’s essentially a new layer in the memory hierarchy that reaccelerates the system at a faster pace, post-hibernation.