INTRODUCTION
I'm an information technology professional with several years of experience. I earned my MCSE early in 1999, as a means to help further my career. I'd been fiddling with Linux at home for only a couple of months, at that time. In the months between then and now, I've learned an awful lot about Linux. I'm no expert - not by any stretch of the imagination - but I'm pretty comfortable with what I need Linux to do.

I should qualify the statement above by stating that I'm not an expert with Windows NT, either. I've got a certificate that says I passed six exams successfully. I do feel competent with NT - more so than with Linux, actually. I see a lot of strengths and weaknesses with both products, and advocate each in their own capacity.

I recently obtained a new job - designing, installing and supporting a network for a medium sized company with _no_ information technology infrastructure currently in place. A network engineer's dream come true: the chance to build my own network from the ground up. I am responsible for all of the technical decision making - desktops, servers, office suites, email packages, security, backups, training. Everything.

As an MCSE, I can spit out a recommended Windows NT domain model in no time flat. One PDC and one Exchange server at the main facility, a BDC at the remote office, and either Windows 98 or Windows NT Workstation on the desktops. However, the organization is a non-profit, and since I have some experience with Linux, I wanted to evaluate the viability of an Open Source solution. Further, the licensing costs of Microsoft products can be nicely side-stepped with a Linux implementation.

CLIENTS
Although I have experience with Linux, my users don't. And I don't have enough experience to feel comfortable supporting 50+ users with things like X Windows (which I can barely get to run myself!). So I'm fairly adamant about installing a Microsoft operating system on the client desktops. These days, most users have a PC at home, so the familiarity of Windows 9X, or the similarity of Windows NT Workstation, works in my favor.

If I had chosen a Linux solution for end user desktops, then I would need to evaluate an office suite. The bulk of my users will be using a word processor, a spreadsheet package, some email, and the occasional presentation. StarOffice 5.1 comes to mind as a free, robust, capable package. But the catch is that StarOffice is only free for non-commercial use. Since we're a business, we're still obligated to pay for licenses to use the software.

If I'm going to pay money to use a software package, then I want to ensure that there are ample support resources available to me in order to support my end users. If someone experiences some bizarre anomaly with Microsoft Office 2000, I can quickly consult TechNet, or the MS Knowledge Base, or ask any number of people (professionals and lay people) who use MS Office regularly. If an issue arises with StarOffice, where do I turn to for in-depth support information? How long will it take me to get comparable answers? How long will my users have to suffer downtime while I chase down the limited support options for StarOffice?

Further, Microsoft makes the Office Resource Kits available, allowing me to really customize my Office installations, as well as provide me with some helpful troubleshooting tools. To the best of my knowledge, no such thing is available for StarOffice.

And finally comes the issue of training. If I were to install StarOffice (or any other Linux-based office suite), who can I hire to conduct professional training sessions for my users? I can only think of one place in Columbus, Ohio, who might be able to accommodate me on this, and the lack of competition suggests that pricing may be higher. Otherwise I'd have to conduct in-house training for all of our users. And if I'm busy training users, how much other work am I free to get done?

NETWORKING
I was familiar with Samba from my own Linux network at home, but only for workgroup based sharing. I eagerly started learning as much as I could about Samba, and its capacity to act as a Windows NT server. The connectivity itself was easy enough to implement. After a couple of false starts, I was quickly using my Linux test server to authenticate domain logons from my laptop. I configured the Linux box as a WINS master browser, and network browsing was working just fine.

My first real snare came when I tried to change my network password from my Windows 98 laptop. The change would never take. I did a little more researching (scoured the included Samba documentation, asked my local users' group, etc) but found no evidence that password changes from a Windows 98 client were supported. I'm told that Windows NT Workstation can accomplish this necessary task, but I have not tested it myself.

After a little more poking around, I realized that Samba had no capacity to expire passwords. The Linux logon account would properly expire if I were to try to log on from the server console, but Samba happily authenticated expired accounts without complaint. This is, quite simply, unacceptable. One of the main components of network security is password security. One component of password security is mandatory password expirations. If my network can't require users to change their passwords every 45 days, then passwords are nothing more than a formality.

I began to toy around with the idea of implementing some sort of kludge - a counter in the logon script, maybe - to get the Linux system to force a password change. But the more I thought about it, the more concerned I became at the amount of obfuscation involved with such a process, and the fact that it could foul up. The Linux user account had one password, and the Samba account had a password. They should normally be synchronized, but the fact that there was a possibility that they could become disparate was bothersome. This meant that a user could, for example, authenticate a logon but be denied email access. Or vice versa, conceivably.