Why They Die
Dec 10, 2008 07:01 PM
79,874
Enter your username and password.
Some say that the end of the trusty hard drive is near, killed by SSD. But let's not be so quick to give up on a technology that stores a whole terabyte for $100.
It'll be years before solid-state flash-memory disks (in this case usually referred to as SSDs) let us cheaply bank the same amounts of data as trusty old hard disk drives for a reasonable price. So, you might as well know how they work, 'cause honestly, they'll have a place on or next to your desk holding all the crap that won't fit on daintier solid state drives—HD movies, huge pictures, music and who knows what else if you're Jason Chen.
What Goes on Inside
The reason hard drive is abbreviated as HDD is that it's really a hard disk drive. Inside you've got what's called a "platter," which is a magnetized recording surface that spins around really really fast, with a head that zooms across the disk to read and write data, think kinda like a record player, except that the head never actually touches the disk except, as you will see below, when bad things happen. [Image via Wikipedia]
Hard drives also come in a few different sizes, with 1.8", 2.5" and 3.5" being the most common, but they've been bigger (and smaller). 3.5" is for desktops, 2.5" is for notebooks (or obsessively quiet desktops), and 1.8" is what goes in classic iPods, MacBook Airs and other small portable devices.
The more platters a drive has, the more data it can hold, but most advances in storage have focused on increasing storage density. A really high-capacity drive can have four platters, while many 3.5" desktop models and some elite laptop 2.5" drives have three platters. Most laptop drives and all the 1.8" portable-device drives that we know of are limited to two platters.
The real catalyst for those 1TB and 1.5TB monster drives pooped out by Hitachi and Seagate wasn't platter stacking, though. It was perpendicular magnetic recording, which allows for triple storage density by storing data vertically (or perpendicularly) along the platter's recording layer, rather than spreading it out across it horizontally (parallel-ly?). However, data is more fragile and susceptible to erasure when stored vertically, hence the slow creep in precision allowing for greater storage densities and capacities.
What All Those Numbers and Letters Mean
You might've noticed hard drives are often labeled as IDE or SATA or PATA or PITA (kidding), with specs like 5400RPM or 7200RPM, plus they come in various sizes, like 1.8, 2.5 or 3.5-inches. Confusing, no? So here's all that crap means.
RPM means the same thing it does in cars, rotations revolutions per minute. In hard drives it's important because the faster the disk spins, the faster it can read and write data. 7200RPM is the standard for desktop drives, but performance models run at 10,000RPM or 15,000RPM. Notebook drives typically run at 5400RPM, because they're smaller, but recently, you can order them with 7200RPM to get more performance at the cost of battery life.
A higher RPM is the single greatest performance variable, since the faster it spins, the more data it can read or write within whatever time frame—it also makes access faster, since the head doesn't have to wait as long to pass over the right data once it's moved to the right spot. And a faster (lower) seek time, basically, refers to how long it takes for the drive to move its head where it needs to go to read or write data. High end drives have a seek time of just 2ms, while typical consumer drives are close to 9ms. Also, the higher the buffer—most typically 8, 16 or 32MB—the more data it can pre-cache, though Tom's Hardware found that you getdiminishing returns there.
How They Connect
The various kinds of drives essentially refers to how it interfaces or connects with your computer's motherboard. There are a bunch, but only a few worth knowing. Up until the last few years, the dominant standard was ATA, or Advanced Technology Attachment. Once SATA, or serial ATA, came onto the map (more on that in a sec), regular ATA picked up the alternative name parallel ATA.
Further revisions to the ATA spec allowed for hard drives with greater storage and faster transfer speeds, and you might see drives using the later spec revisions called "Ultra ATA" or something similar, and they can transfer data at 133MBps (which is slooooow). ATA drives are commonly called IDE (integrated drive electronics), but ATA is more precise. If you've ever messed around inside a computer, you'd recognize them because they connected to fatass ribbon cables that take up a lot of room. The third major interface, which you should know of, but not necessarily about, is SCSI (pronounced "scuzzy"), which was primarily used in the enterprise or high-end space when ATA was still king. The ATA/IDE interface also confused some with its master/slave assignations, which, as you'll see, is no longer a problem.
Okay, so the current hard drive standard in consumer PCs as of a few years ago is SATA, which is worlds better than ATA. For one, it's faster—first-gen devices ran at 1.5Gbps, but now they're up to 3Gbps, and are on the road to hitting 6Gbps. Also, their cables are way thinner, for better air flow and less tangly crap inside your case. And because they're smarter and don't depend on a lot of configuration, they're easy to work with, and are even hot-swappable. Newer external drives use a variant of SATA, eSATA (e for external) that essentially just moves the port to the outside of the computer case, delivering SATA speed for peripherals. Soon, eSATA will come in a bus-powered format, much like the smaller portable USB drives you see today.
Fast seek times are different than fast transfer times from a good interface—one pertains to how quickly the data can be located on the disk, and the other is how fast it can be sent over. To describe it in somewhat oversimplified terms, you can see how a slow interface on a fast seek drive would be better for a system that's constantly shifting tiny bits of data, where a fast interface on a relatively slower drive is good for moving really large files around.
Why They Die
Remember how I said the head usually never touches the drive's platter surface? When the head actually does touch the drive platter, it's what's called a head crash (check out the video above), and it means you're skee-rewed. Normally the head flies on a tiny pocket of air, but a single particle can make the head bounce on the disk, totally hosing the magnetic layer, especially at higher RPMs. And it just gets worse from there, because stuff scraped away by a head crash making it more likely that more head crashes will happen. More mundanely, the delicate mechanical parts eventually just wear out over time, which is typically measured by the the drive's rated mean time between failures. Unfortunately, there's not a whole lot you can do to predict when your drive is gonna go down in flames, unless you bought a drive from a series suffering manufacturing defects.
So what is really the single most important thing you should know about hard drives? Back your crap up, they may be awesome, but that doesn't mean they're without weakness.
Something you still wanna know? Send any questions about drives, personal storage or other hard things to tips@gizmodo.com, with "Giz Explains" in the subject line.
Send an email to matt buchanan, the author of this post, at matt@gizmodo.com.
Original material is licensed under a Creative Commons License permitting non-commercial sharing with attribution.
Start a new discussion