This article purports to talk about the cost of using Linux in the "enterprise" as oppossed to Windows or Unix.

I'm naturally suspicious of any trade article that talks about "real cost" and is buzzword laden (TCO, ROI etc. etc.)

Probably the most honest line in the article is the following:

    Figuring out the TCO of Linux is not for the
    faint of heart. Several IT execs told us they
    skipped the exercise because the model would
    have been very complex, with too many unknowns 
    and assumptions.

Substitute [your favorite bit of software] for [Linux] and you are getting closer to the truth. TCO and ROI are business school buzzwords that are almost impossible to measure in the real world.

Nonetheless there are some interesting "rules of thumb" that emerge in the article, that can help in the decision making process:

1. Linux software is free. The main cost involved is staff training. This is an up-front investment, which when amortized over time is probably negligible.
2. Urban-myths to the contrary, maintenance costs of Linux are far lower - less patches to apply, less downtime, greater stability - all mean less staff needed.
3. If you are already a Unix shop, staff training is a very minor issue.
4. In terms of reliability, efficiency and safety, Linux wins hands down over the Microsoft alternative.
5. The biggest barrier to migration are those pesky proprietary apps that are tied to Windows servers, and to a lesser extent to Unix.

In short, in terms of any meaningful measure, it pays to migrate to Linux. It's that proprietary vendor lock in that is holding you back, making you spend far more and get far less. The obvious lesson is this:

In the future, stick to FOSS for as many applications as you can. Avoiding vendor lock-in is the best way to keep IT costs down and maintain an efficient and safe IT operation.