Last Updated on March 23, 2022 by Dave Farquhar

I started my professional career doing network administration at the University of Missouri. (I generally don’t count my stint selling low-quality PCs at the last surviving national consumer electronics chain towards my professional experience anymore.)

The University had its own IT department, but some of the larger departments, particularly Journalism, had their own IT departments as well. I worked for the School of Journalism. The School of Journalism had one of the oldest Token Ring networks in the world. It was also home to the oldest OS/2 network outside of IBM itself, dating to the late 1980s, running on pre-release versions of OS/2 and pre-release hardware. Some of the pre-release IBM PS/2 Model 80s survived in production until 1998 when I decommissioned them. They don’t make ’em like that anymore.

Windows NT began life as OS/2 3.0. Although IBM and Microsoft soon stopped collaborating on OS/2 and went their separate ways, from a networking standpoint, OS/2 and Windows NT were highly compatible. By 1997, it was clear that OS/2 wasn’t going to meet the school’s needs for much longer, so we made the decision to replace the OS/2 servers one by one with Windows NT servers, eventually ending up with a Windows NT network. I was in charge of the project.

One day, all of us were summoned to a meeting. The campus had several Windows NT gurus who, while knowledgeable, were also extremely anal retentive. The meeting was to enforce new naming policies. All networks had to be named UMC-something.

“Some student set up a Windows NT domain in his dorm room named Barfy,” said the loudest, most annoying and most anal-retentive university administrator.

We didn’t like this policy. Our Windows NT network was named MUJournalism and consisted of hundreds of PCs. What’s worse was the network was extended out to the university-owned television station on the edge of town, several miles away. Renaming our domain to fit these guys’ whims was going to be a lot of work for no benefit whatsoever. Also, as I recall, there was also some technical reason why the name UMC-Journalism wouldn’t work. Perhaps the name was too long for our remaining OS/2 clients and servers to handle.

And besides that, our network had been named MUJournalism, or some variant of it, since the last days of the Reagan administration.

We didn’t change the name of our domain.

We did, however, take a handful of test servers and set them up in their own domain. Our Lotus Domino administrator/programmer chose the name: UMC-Barfy.

A year or so later, I was working at my second employer. We had pockets of departments running Macintoshes, some of which were nearly as old as those old PS/2 Model 80s and roughly as dependable. To reduce acquisition and support costs, we were replacing as many of those as possible with Windows PCs made by Micron.

The Windows NT administrators at this place were less than accommodating. I needed some way to get the data from these Macintoshes onto the new PCs. Popping the drives from one machine into the other wasn’t an option–Windows NT wouldn’t read the Macintoshes’ HFS and HFS+ file systems, and the Macintoshes wouldn’t handle NTFS, the mandated standard. But besides that, the drives in the Macs were SCSI, while the PCs were all IDE.

Easily the fastest and best way to move the data would be to bounce it off a file server. Windows NT’s Services for Macintosh wasn’t the most reliable thing in the world, but it was adequate for a job like this. So we requested that Services for Macintosh be added to one of the Windows NT servers in the building. Temporarily of course.

Our request was denied.

We explained that this was necessary for a migration that was happening with great encouragement from upper management. Upper IT management, at that.

The request was still denied.

My boss happened to have an unused copy of Windows NT Server in his cubicle. Needing to get this done, we took an old consumer-grade HP Pavilion PC that was too old and slow to be good for much else and proceeded to install NT Server on it. As we were doing this, I related my story of the rogue Barfy networks.

I guess he liked the story, because when it came time to name the server, he seized the keyboard and typed BARFY for the name.

Windows NT finished installing, so we tucked Barfy into a corner and I proceeded to finally migrate my first Macintosh.

It was shadow IT. But sometimes shadow IT happens because IT pros are being jerks.

The next Monday, the crankiest of our unhelpful Windows NT administrators tapped on my boss’ cubicle wall. “Do you know anything about a server named… Barfy?”

He waved his hand. “This is not the server you are looking for.”

Unfortunately, Jedi mind tricks don’t work on Lutherans of German descent from Wisconsin. Or at least they didn’t work on this one.

So the two of us got our hands slapped–something which became a yearly tradition, at least for me, until this guy left for greener pastures a few years later–and he made us unplug Barfy from the network.

We were in trouble for a lot of reasons, but the biggest reason was because you weren’t allowed to build Windows servers unless you’d passed the Wolfe Aptitude Assessment Battery Programming test from 1968. My boss and I had only passed the Wolfe Computer Operator Aptitude Test from 1968. This was 1999 and I wish I was joking about the tests. Which of the two inappropriate tests was less incorrect for this use case was a constant matter of debate when I was there.

So I commandeered a cart, a couple of power strips, an old 3Com 10-megabit hub, and some network cables. Migrating a Macintosh became a matter of wheeling Barfy into the cubicle, unplugging the Mac from the building network, plugging the Mac into the hub along with Barfy, logging in, and copying all the user data up. While that was going, I would plug the PC into the same hub, log into Barfy, and then copy all the data back down. Then I would unplug the PC, plug the PC into the office network and reconfigure it, and haul off all of the old Macintosh equipment and put it in a pile.

It wasn’t very efficient, but it kept the uptight Windows NT administrators happy and it kept their servers in the state they wanted them in.

And I guess it gave me a chance to act a little like MacGyver.

If you found this post informative or helpful, please share it!

4 thoughts on “Barfy.

  • September 23, 2009 at 9:49 pm

    I’m not sure it was so much the naming we used – that was probably bad enough in itself. It also seemed that Compaq blamed Barfy for bringing down the DC because of competing Mac services on the network or some such mularky.

    I never thought it was a bad thing about Barfy. If nothing else, it seemed to open the opportunity for a Snap server!

    • September 25, 2009 at 8:27 pm

      I forgot that little detail, although I still fail to see how competing Mac services can bring down a DC seeing as there WERE no other Mac services running in the building at the time. But we’re talking the same guy who got onto us for using Ghost initially. (I think that was my first scolding… Can’t remember now what the third and fourth thing was, but I never went a year without doing something that needed to be done that he didn’t like.)

      I agree, broadening their horizons a bit was a good thing.

  • September 24, 2009 at 9:11 am

    This is exactly what happened to me with my "Windows NT 4.2 Server" (which is what Samba announced itself as, if I remember correctly, when installed using MkLinux on a Power Macintosh 6100/60). Right down to the cranky attitude.

    You were gone before my most impressive user-visible network changes, though, like the time I went temporarily insane and misconfigured our colocation switches and brought the whole place down for a couple of hours.

    That reminds me… LD left us a couple of weeks ago.

    • September 25, 2009 at 8:32 pm

      The whole colocation facility, or just your little segment?

      As for LD… You’re trying to lure me back, aren’t you?

Comments are closed.

%d bloggers like this: