Given all the emphasis on TCO lately, you probably feel that you can exercise control over it. The truth is, if you think TCO stands for Total Cost of Ownership you're sadly mistaken. Acquisition costs, although only a portion of the TCO calculation, are a major and immediate in-your-face expense. The knee jerk reaction is to try to modify the shopping list, but often the best solution is to spend in order to save over the long term--acquisition costs are generally not subject to laws of fungibility.

For instance, a server is monumentally more expensive than a desktop system. Start at the CPU: By popular rumor, Intel sells a 64-bit 800MHz Itanium server processor with 4MB of cache for $4,227. On the desktop side, Intel's 32-bit 2.2GHz Northwood Pentium 4 with 512K of cache sells for $654 retail. Think about it: The desktop CPU moves only half the bits of the server CPU, but runs at nearly three times the speed and costs about 1/7 the price. But you can't put a desktop CPU in a server. It's just not done.

And what would a server be without a SCSI hard drive? Drive prices solicited from CDW's Web site revealed that a Maxtor 18.3GB Ultra3 SCSI drive runs $179, whereas a 20GB Ultra ATA/100 drive costs $106.72; and prices scale accordingly on up the line. This is not a swipe against either Maxtor or CDW; such disparities run across drive brands and resellers. Worse still, much of the data you see about how much faster SCSI drives are compared to IDE devices is mostly hype; the transfer rates you'll see for both types of drives are taken between the buffer and the host interface. That's like bragging about an incredible golf shot you made on the moon. It's just not real world.

So how do you find the spots where you can shave dollars off your TCO? The answers can often be counterintuitive. Like any good IT manager, you don't cut corners, you just round them off a bit. That means looking at what's available in relation to the best long-term results. For example, blade servers are initially more expensive than traditional servers--normally a huge red flag for conventional TCO evaluation. But a blade server can provide savings as great as 30 percent over traditional servers if you consider the cost reduction attached to reduced square footage and power consumption. As well, a dense server farm that shares more components lowers your maintenance costs over the life of the equipment.

If you're feeling particularly out of control at this point, there's always your support staff to consider. It's typically one of the first targets when the axe is aimed at the bottom line. Unfortunately, if your productivity declines because of increased downtime, your real TCO actually rises. And hopefully, by now, you've begun to get the hint that TCO actually refers to Total Cost of Operation.

As for user support, you should increase that, too--but not in traditional terms. Body counts bump TCO up. Instead, automate. A remote management policy allows one person to cover multiple incidents in less time than several people could. Whether it's a trip down the hall, in the elevator, or across town, travel time is dead time and time is money. And therein lies the myth: TCO isn't a number that appears on the bottom of a shopping list; it's the cost of doing work over time leveraged against your productivity.

What's your experience with acquisition costs and TCO? E-mail Bill or post your thoughts in our Talkback forum below.