Happy 25th birthday to the PC! In early August 1981, IBM unveiled the IBM PC. Although many argued that the PC was built from a lackluster array of parts that added up to an unimpressive whole, the fact that Big Blue had weighed in on the microcomputer world meant a lot: Buyers could benefit from a set of software and hardware standards in a time when standards were hard to come by. Within two years of its release, the PC had turned things such 360KB 5.25" floppy-disk drives, the ST-506 hard-disk interface, and the PC bus into standards. Standards that a big company has backed are the seeds of a fast-growing market: Choosing components compatible with those standards is less risky, and prices inevitably drop.
Okay, you knew that part. Let's talk about the downside of standards. When you've got a set of good hardware and software standards, innovation starts to slow down at a certain point. A quick look at prices in August 2006 shows me that I can buy a 2.5" (notebook-sized) hard-disk drive with 160GB capacity for about $200. I can recall paying $929 for a 32MB drive back in 1985, so this seems to be a good deal. Progress marches on, right?
Well, maybe not. Cheaper drives are good news, but the news could be better. Once upon a time, CPUs communicated with their RAM at full processor speed. That hasn't been true for years. Data-transfer rates on hard disks are faster but not much faster than they were ten years ago; most notebook hard disks are 4200rpm or 5400rpm rather than the 7200rpm offered by a few—and I remember those speeds being common even 15 years ago. Yes, you can buy drives with rotational speeds in the 15,000rpm range, but not for the inexpensive IDE family of drives, and I’m not aware of many notebooks that run SCSI drives. (That's important because people have purchased laptops more often than desktops since May 2002.) Hard disks were fragile in the old days and, unfortunately, they’re still fragile. I often reflect on what must be literally petabytes of vital personal information sitting on home computers, and I shudder when I think about how little of it is backed up. Storage, to take just one example, is still a pain.
How can the situation get better? I've got a couple of suggestions. They might require mildly new standards, but I think it's about time for a little shaking up, don’t you?
First, let's make drives more reliable. For about $70, you can buy a 2.5" hard disk that stores 80GB of data. The disk drives aren't physically large&$8212;roughly the size of four stacked credit cards. So, let's take three of them and add some electronics so that they act as one RAID-5 array. The computer would recognize the result as a 160GB drive that’s pretty fast on reads and that’s able to withstand some physical-damage scenarios that would be impossible with a single-drive storage approach. The interface could resemble a standard ATA-6 interface and so be hardware-compatible with current standards. Add a few extra commands to retrieve RAID status information for health-monitoring purposes, and you've got an interesting, compact, safe storage mechanism. (The laptop hard-disk bays would have to be enlarged, and cooling might bring new concerns, but we're busting standards here.)
Here's another possible new standard. Why not exploit the now well-understood power of fiber optics to create a replacement for the FireWire and USB standards? Fiber can support terabits per second, and optical cables are no longer horrendously expensive compared with copper. Create an interface that can handle terabits per second over distances of a few feet, and suddenly inexpensive external drives such as the ones we currently see with USB and FireWire interfaces—which traditionally are far slower than internal drives—become indistinguishable performance-wise from internal drives. Set up the BIOS so that you can boot from an external "FiberWire drive"—cool name, huh?—and it's a snap to choose to boot from a half dozen different OSs on a given machine. (After all, virtual machines can't do everything.)
The PC revolution changed the world 25 years ago. But that’s ancient history now. I fear that the business has become a bit hidebound. Let's start talking about what PCs still don't do that we've been waiting for them to do for 25 years. Then, we'll have something to celebrate at the PC's 50th!