To answer that question and a slew of others, I sat down today with Steve Weinger, senior marketing manager at Samsung Semiconductor, and picked his brain about all things SSD. Here's what I learned.
SSDs are made up of chips of Flash NAND memory — the same chips that are used in the RAM on your computer, or the memory inside your cell phone or iPod. These chips can be arranged in all kinds of stacks and shapes, but for now, most SSDs occupy the same form-factor as traditional 2.5-inch or 1.8-inch hard drives, since most notebooks are already designed around these sizes. The Flash chips that make up an SSD drive are all alike — "basically a commodity," as Weinger says. Hang onto that tidbit of knowledge; we'll come back to it later when we talk about those big upgrade prices.
SSD chips can be of two varieties: multi-layer cell and single-layer cell. SLC chips are more expensive than MLC chips, which contributes to the big price-tag on SSDs when you order your note book built-to-order, since most of the SSDs in laptops today are SLC. However, SLC doesn't have much of an advantage over MLC; its only edge is that it lasts longer. Important? Not really, when it comes to these drives. SLC drives can last about 20 years, and MLC over 10. Show me a notebook that lasts longer than 10 years, and I'll show you a notebook that's as good as a paperweight. Traditional hard drives last about 4-6 years.
Samsung and a handful of its competitors have decided wisely that MLC drives are the way to go. While most companies are just getting around to putting out 64GB SSDs (ahem, here's looking at you, SanDisk), Samsung has already announced 128GB and 256GB drives will be available to OEMs by the end of the year. Rumors also abound that a Samsung 320GB isn't far off either, but Weinger wouldn't confirm or deny that.
So what about power savings? There have been some rumors circulating on the Net that SSDs are actually less efficient than traditional hard drives, which contain platters of disks that are read by heads (kind of like the way old record players read LPs.) Weinger says that information is inaccurate, because the test that was cited in those rumors was performed on old SSDs. Get ready for a lot of acronyms: old SSDs used interfaces called SATA1, which were essentially old PATA interfaces with a bridge converter. That converter took up a lot of juice. Nowadays, modern SSDs are SATA2, which don't need a bridge and have twice the throughput (3Gbps instead of the old 1.5Gbps.) So never fear — SSDs are indeed more battery-friendly.
So if they save power, that's great, but why else use them? For one thing, your computer can access files much more quickly on a solid state drive than it can on a platter drive. That's because the machine doesn't have to wait for a head to search around a stack of disks; the machine goes directly to the source of the file. That results in dramatically quicker startup and search times, and has another added bonus: no deterioration of speed over time. When you fire up a traditional hard drive for the first time, it can be lightning quick, but in a year, it may seem dramatically slower. That happens because the drive needs to move things around and hunt more carefully as more data accumulates on the palettes. No such problem with an SSD.
So how prevalent will these things become? Weinger doesn't believe they'll ever replace traditional hard drives completely, but they are poised, he argues, to gain an incredible amount of marketshare in the next three years. By 2010, he says, the majority of all computers sold will be notebooks, and about 30-40% of those notebooks will contain SSDs by 2011. That's pretty bullish, considering that only about 1% (or less) of notebooks purchased have the drives today.
Contrary to their present market — ultra-thin gadget loving gurus who buy MacBook Airs — SSD-equipped notebooks will see enterprise customers as their primary early adopters, says Weinger. Companies will be attracted to the drives' vastly superior durability and reliability, since time and work lost to hard drive failures is a significant cost to many businesses. Companies will also begin to employ certain kinds of Flash-based SSDs in their servers to do what is called "short stroking" — quick data access, as opposed to mass storage, which will likely stay on traditional hard drives. Those drives will be chosen also for their durability, but primarily for their speed and power savings. Cooling a server room is shockingly expensive due to the amount of heat that servers generate, and since SSDs offer about a 90% wattage reduction over traditional drives, green (and fiscally prudent) companies will want to incorporate SSD servers in their racks simply for the vast energy savings.
So with 30-40% of notebooks sporting these drives by 2011, and businesses clamoring for servers and notebooks including them, will the prices go down? Yes, says Weinger, but more importantly, they'll stabilize. Since Flash chips have become a commodity as I mentioned before, their price fluctuates as any commodity does, in response to the market. Since most Flash-based devices right now are low-priced consumer electronics like iPods or digital photo frames, Flash demand (and therefore, price to manufacturers) goes on a roller-coaster ride every year that tops out around holiday time and lulls in the summer. The more enterprise and consumer computing hardware that requires flash, the more consistent the demand will be, and in turn, the thinner the margins can be on an SSD-equipped computer without the manufacturer worrying that they'll take a hit if the raw material (ie, those Flash chips) unexpectedly surges in price.
Whatever their use, Flash chips are one of the most promising evolutionary improvements being seen in the computing industry today, and I personally can't wait for their ubiquity. Anyone else whose had a hard drive failure in the last few years will probably join me in that sentiment. Luckily, our prospects are looking good.