I read more and more often that PC hardware vendors are bemoaning their lack of sales growth and wondering how to inject more excitement into the business. In this month's Windows & .NET Magazine Update Special Edition, I attempt to illuminate the problem and provide a solution. (I limit this discussion to server hardware, but I can think of plenty of examples that apply to desktops, as well.)
The underlying problem stems from people like me. I'm cheap, I run a small shop, and therefore I don't need 16-way systems. (That doesn't mean I don't want them.) When I shop for a new server in today's market, I can buy an adequate new system for about $6000. But I'm always excited by the possibility of acquiring some cool new piece of hardware in the process, which tends to subdue my more parsimonious self.
High-end systems do have some goodies that are untouchable in my price range. My helpful suggestion to hardware vendors, then, is to find a way to move these innovations into the less than-$10,000 systems. (And that means that one day they'll make their way into the $1000 systems. But I'm not greedy; I can wait a while for that day.)
Take, for example, hot-swappable PCI. My Windows 2000 servers rarely require reboots--only after hotfix, service pack, and hardware installations. Microsoft has promised that more and more hotfixes won't require a reboot, but I don't know that we'll ever see systems that don't require reboots after you apply a service pack. But hot-swappable PCI--gosh, I could like that. For example, I have a FireWire II card that I want to install on my Web server. I've had the card for a month now, but I have no idea when I'll be able to install it because I hate creating downtime on my Web site. Yes, I could set up a cluster as a way to upgrade hardware without downtime, but clusters are still expensive. So I look forward to the day of the $1000 server with hot-swappable PCI.
But hot-swappable PCI technology doesn't speed up my systems--it just makes them more reliable. Reliability is certainly good, but it's not exciting. What exiting features do the buyers of expensive systems have that we frugal types lack? Processor architecture.
As I recall, the Intel 32-bit processor architecture that we base our systems on turns 18 this month with the anniversary of the 80386DX's release. Sure, the 486 and Pentium processors improved on the original 80386 in some ways, but the opcodes aren't that different--and when a processor architecture reaches voting age, you know it's time for one of those paradigm shift things. That leads to my next suggestion: Crank out those Opterons.
AMD's much-anticipated 64-bit Opteron processor is set to debut next month. It should be an impressive new CPU but, as usual, pretty pricey for a few years. We all know that the new CPU will eventually drop in price--to a point at which you could cost-effectively use it to run a programmable TV remote. The key to increased sales is to accelerate the price fall.
We need a large semiconductor firm to announce the imminent release of a new 128-bit processor, slap together some believable technical documents, and hold a press conference. In no time, we'll see both the Itanium and Opteron processors drop to price levels that all but the most impoverished can afford.
I have plenty of other suggestions to save the hardware business, but I've run out of space. I hope hardware vendors will see the logic of my plan and that you'll be reading next month's commentary from your new OpteronSL-based laptop--the one that offers 7 days of battery life.