The upgrade that wasn't

This is a story about a vertical market application that was running on a SCO system. Yes, that SCO.. the one that's just about out of business now and, in the minds of many of us, are reaping just what they deserved. That SCO.

I know about this particular application because I have a customer who uses it. Over the past few years I've helped them out with various SCO and Linux issues (they have a Kerio Mailserver running on Suse also) and yes, even some Microsoft Windows issues. I know the owner pretty well; he's a really nice, hard working guy and absolutely no fool when it comes to technology.

About a year ago my customer told me he was going to have to switch to Windows. He didn't like it, but his application vendor told him that they would no longer upgrade the Unix version and that new features he really wanted would only be in the Microsoft version. I had written some helper programs that had provided some of the new features he needed, but of course I couldn't integrate them fully with the application. He really would be better off with the upgraded application, so he bit the bullet and told them to go ahead.

This required a $5,000 Windows Domain Controller. Of course I protested that this was not absolutely necessary and referred him to my "Do you really need a domain controller?". However, I also warned him that if he did not provide their "recommended environment", they'd be apt to use his non-compliance as an excuse for any glitch that came up.. often its better to just go along to get along..

Of course there was also a six or seven thousand dollar software upgrade which included the application vendor coming out and doing installation and training. Before they got to that, though, there were problems with the domain controller: I don't remember whether it was bad hardware or flaky drivers, but it caused a fair amount of grief immediately - added to the normal confusion involved in switching Windows users from a peer to peer environment. But that got fixed, and the installation and training people eventually showed up, did their thing, transferred data, and started the training sessions.

This went badly.

As my customer explained it, on the old Unix system they were all accustomed to using Facetterm and had eight Facet windows open for each user - this let them look up related information extremely quickly. There's nothing quicker than a Ctrl-W, especially when someone has been using the system for a while and knows exactly what window they want. Watching an experienced user flash through text windows at incredible speeds should make anyone realize that point and click is hopelessly inefficient and slow.. and of course the users complained. More annoying was that they couldn't have eight windows anymore - as cluttered and difficult as it would have been to use them in Windows, the sad fact was that their desktop hardware didn't have enough muscle to open more than a handful.. the application vendor suggested buying new desktops with lots of RAM.

The owner wasn't happy. The users weren't happy. Most importantly, the owner's wife wasn't happy and she was the one responsible for most of the really important input. However, everyone agreed to gamely "give it a chance" - they'd keep up entering data in the old SCO system and "play" with the new system. The application vendor probably assured them that their comfort level would grow with experience.

It didn't.

I talked with my customer this morning. He told me that he'd given up, because if his wife had to use the whiz-bang Windows system, she'd quit. That wouldn't be good for the business or his home-life, so he is throwing away his upgrade investment. Well, not entirely: his mailserver could use a hardware boost, so we talked about overwriting Windows 2003 Server with Suse or Fedora - the hardware is way beyond what he'd ever need for a mail and web server, but using 20% of it is better than not using it at all..

Am I happy he's back on SCO? No, of course not: I'd rather he were on a Linux box, but that's not possible. His application vendor is moving on with Windows, telling him that they will be soon rewriting that using Microsoft .NET (and thus typing themselves even more irrevocably to Microsoft technologies). For the moment they will still somewhat reluctantly support the Unix version.. when that stops, my customer says he may limp along without support if he can.. or he'll have to look for other software or spend that upgrade money all over again.

I wish application vendors were smarter about this. Windows is NOT where they should be focusing. That's bad for them and their customers both in terms of overall cost and lock-in to a dangerous environment that will only become more costly in the years to come. Microsoft is going to need more and more new revenue and they'll suck it from those who are foolhardy enough to put on the golden Windows handcuffs. Moving to open source Linux is better all around.