Opinion on TCO: Fuzzy Math, Fuzzy Logic?
Cost of a fully loaded Windows 2000 server box: $5,000
Cost of three years worth of maintenance and support: $150,000
Seeing the looks of glee on end-users' faces: Priceless.
If business were as simple as a credit-card commercial, no one would have to sweat over how effectively his or her IT budget dollars were being spent. But things get murky as you start to price out systems costs, and that's why we have analyst firms to tell us the bad news. Things are never as cheap as the vendors tell us. But what are the analysts telling us?
Gartner has been harping about its total-cost-of-ownership numbers for years, which consistently get washed away by market pricing shifts. Now, the analysts are attempting to put a real price tag on Linux ownership numbers – contrasted against Windows – and it looks like nobody agrees on anything. IDC says Linux costs more than Windows over the long haul, and Robert Frances Group (RFG) says Windows costs more than Linux. (When analyst firms do start to agree on something, we'll all be in trouble.)
TCO, like its more-favored sibling, ROI, are more like the formless clouds used to represent the Internet in PowerPoint presentations. Everyone needs those numbers to justify their spending. Often, however, numbers are arbitrary, and vary from situation to situation. ROI and cost of ownership are precise measures in finance and accounting circles, but are more an art form in IT circles. As a result, the calculations simply become some vague up-front guesswork to get the technology budget past the corporate bean counters. Better to show ROI first, but if you can't, show a favorable TCO.
That's not a bad thing, except that these inexact measures often slingshot back against IT when things don't pan out. IT gets blamed for spending all that money on a system nobody wants. Many implementations simply don't work out because of vague user requirements, changes in user requirements, or changes in the overall business midstream through a project. The challenge is in getting the end-users to use the systems once you get them implemented. Plus, even if new revenue does start flowing, it may be difficult to quantify how much is a direct result of the new system. Data warehouse guru Bill Inmon described the dilemma in a way that could apply to any type of system: "There are many factors that contribute to increased market share and increased revenue, not just the data warehouse. Trying to measure the effect of just the data warehouse on those variables is like trying to pick out a single violin from the sound made by an orchestra."
When it comes to core data center projects – such as migration to another version of Windows – of course, most analysts will tell you that ROI is not possible. Instead, TCO becomes the reigning acronym. How do you put a price on deploying one type of system versus another? Let's face it, any new system implementation, even if it is to Linux, is going to raise your cost of ownership, at least initially.
I know of a pension management firm that has been making piles of money in recent years, in spite of any economic downturn. New accounts keep rolling in. Amazingly, most of this firm's mission-critical operations are still run on DOS-based PCs. The firm is finally considering a move to Windows 2000 Server. What will the ROI be from this move? Negligible, unless you can measure the benefits of reduced employee frustration after moving off the ancient, creaky system. What will the TCO be? Unfortunately, the cost per user will be much, much higher for the roomful of new PCs, networking equipment, and software, not to mention the start-up and retraining pains. But this company needs to move into the next century for many obvious reasons that you can't put a price on. Businesses need to be positioned to begin to take advantage of Web services, for example. And, even if this firm made the move to Linux instead of Windows 2000, the new implementation will still incur the same kind of IT support bills.
Both the IDC and RFG reports make the same point – that software and hardware are basically irrelevant as a cost factor. (Except when you include commercial Unix – then hardware costs go through the roof.) The issue boils down to support costs. How much does it cost to pay someone to build, maintain and troubleshoot the system? IDC says Windows talent is cheaper over the long run, RFG says the opposite. Regardless of who is right, one can safely assume that the real meat in any cost of ownership calculation for Windows or Linux is in the ongoing labor costs. The analysts can't agree on whether Linux or Windows is more expensive – maybe it really doesn't matter anymore.
Joe McKendrick is an independent consultant and author specializing in surveys, technology research and white papers. He's a contributing writer for ENTmag.com.