Usability enhancements or feature regression?

I've always had an insatiable urge to figure out how things work. Whether as a child taking apart my toys or as a teenager writing code, I was driven to answer the question "How did they do that?" I still have the urge to learn how things operate under the hood, and my job analyzing technology and products provides me that opportunity. I still spend a fair amount of time working my way through technology specifications, Internet Engineering Task Force (IETF) Request for Comments (RFC), and software development kits (SDKs).

I believe my need to understand the underlying process is what aggravates me when I work with Windows 2000 (Win2K). In my opinion, Microsoft went out of its way to hide the way things work in too many cases. I realize Microsoft made the changes in the name of ease of use. However, when the features get in the way of my job performance, they aren't truly ease-of-use features.

Here's an example of the problem. Recently, my Internet connection went down. Although losing an Internet connection might not seem unusual, my frame relay service from PSINet had been rock solid for 5 years. Losing my connection didn't worry me too much because PSINet's Network Operations Center (NOC) is responsive and I was sure that I'd be back online quickly. PSINet discovered the problem in less than an hour, but unfortunately, the problem wasn't one that the company could control. The digital line between my home and the telephone company's central office wasn't working (which explained the loss of the signal light on my Digital Service Unit/Channel Service Unit—DSU/CSU), so PSINet had to turn the repair request over to Bell Atlantic to process.

Because I have a corporate account with PSINet, the company wanted me back online as quickly as possible. The support technician asked me whether I had a Windows NT system available (little did he know) and said he would set up a dial-up account for my ISDN line and fax me directions for configuring an NT system as a router for my Class C address. I thought, "Great, this setup will let me try all the new Win2K Internet connectivity features." I had systems on my network running Windows 2000 Server (Win2K Server) and Windows 2000 Professional (Win2K Pro), so I thought the router setup would be simple.

I received six pages of instructions from PSINet about how to set up NT as a router for my network. The process boiled down to four steps. First, I needed to configure DUN to call the access number. Second, I needed to enable IP forwarding. Third, I needed to manually configure the new routing using the Route command at a command prompt. Fourth, I needed to change the gateway address on the network machines that would use the temporary router. (The downside to using static IP is that I had to change the gateway address on all 12 machines in my home network.)

Using NT Server 4.0 or NT Workstation 4.0, I would have needed about 10 minutes to complete the four steps. I would have needed a little longer if I hadn't already installed and configured RAS for the system. Using Win2K, I needed a lot longer. I first tried to use a Win2K Pro system configured for DUN, but the OS didn't provide IP forwarding. After opening every possible Win2K Pro applet related to networking and dial-up, I came to the conclusion that you can't use Win2K Pro to enable IP forwarding.

I made the leap of logic that IP forwarding must be part of the Internet Connection Sharing (ICS) application. I fired up the ICS wizard and discovered that although ICS performs the IP forwarding function, it requires the other machines in the network to use IP addresses in only the nonroutable 192.x.x.x range. Microsoft makes no allowance for multiple valid static IP addresses when you use ICS and a Win2K Pro system for routing.

A quick check of the Win2K Server system revealed no easy way to enable IP forwarding, although employing the router wizard and configuring the server system to act as a temporary router for my network was possible. I called Microsoft support, and the representative told me to use the router wizard. But after wasting the better part of a day trying to get this solution to work on Win2K, I moved my ISDN to an NT Workstation 4.0 system and spent 15 minutes installing RAS and getting the temporary router connection running.

I wasted almost an entire workday because someone at Microsoft decided that Win2K didn't need an option to enable IP forwarding. Exactly how did this decision make my life easier?

You also might have noticed that Win2K reconfigures things to suit the way you work, or at least the way the OS thinks you work. At one point, Win2K determined that my laptop's PC Card modem didn't support hibernation. Therefore, with nary a whisper, the OS disabled hibernation, which left me to figure out why the option was no longer on the shutdown menu.

Win2K's reconfiguration caused a much more annoying problem when I was using a USB-connected UPS. Using USB, instead of a serial port, for UPS signaling is a great idea when the OS supports USB. In many cases, handheld computing device connectors, modems, and digital cameras limit the number of free serial ports. However, UPS needs a computer connection to know when to perform automatic shutdowns and close applications.

The first system I attached the USB-connected UPS to was running Win2K Release Candidate 3 (RC3). The USB tab was available in the Control Panel's Power Options applet, but the OS didn't recognize the attached UPS. The OS didn't present a New Hardware Found message similar to the messages I received when attaching USB devices in Windows 98, so I presumed that the OS wasn't properly supporting USB. Therefore, I decided to perform a new installation of Win2K Pro on a new machine. I began the installation with all the peripherals attached to give Win2K sufficient opportunity to find the USB device.

During the installation, I found no indication that the system was aware of the USB device. The OS took no longer than usual to install, and the OS didn't enumerate the USB port device at any point that I noticed. After the installation completed, I configured the networking options and the screen resolution. Then, I opened the Power Options applet to discover that the OS had automatically configured my Advanced Configuration and Power Interface (ACPI)-compliant desktop as a notebook computer. Also, the applet didn't provide a UPS option tab, and the OS had enabled hibernation. I was somewhat confused and startled. I investigated and discovered that the system thought it was running on battery power. I brought up the Power Meter and clicked the battery icon that showed 100 percent. The configuration details showed that the system had found and identified the USB-connected UPS, then determined that the UPS was a battery powering the system, rather than a backup device.

I don't think that Win2K's tendency to perform hidden operations provides the type of "help" that an administrator needs from an OS, because the options make the product more difficult to understand and use. But maybe I know too much for my own good, just like any other experienced NT systems administrator.