Dear Linux Magazine Reader,
Computer revolutions occur every 3--5 years. Something new happens that changes everything. Of course, we always have the marketing buzz telling us that revolutionary new things happen every 3--4 months, but I'm talking about really significant changes.
A recent revolution that everyone has been talking about for the last few years is the web services revolution. In the "web services paradigm," as it is called, the tools of the web form a framework for building custom applications tailored for the needs of a corporate customer. If you are computer-savvy enough to be reading this magazine, you probably learned about web services technologies a long time ago, and you don't need a refresher course on how web services are taking over the world. Or do you?
I had an interesting conversation recently with a friend who works in the medical services industry creating custom medical records applications for hospitals. This would seem like an ideal scenario for web services: a web server, a back-end database, a little Java code, and some web clients - perfect material for a slide show by a web services evangelist from Sun, Microsoft, or IBM. Or is it?
According to my friend, a couple years ago, all his customers insisted on a web services solution. Every request for proposal included instructions to base the design around web services. But now, he said, his customers aren't paying so much attention to web services. They just want the low bid, and they aren't so particular about where the low bid comes from. And more often than you would think, the low bid isn't the web services bid.
According to my friend, once you factor in the cost of a web server with failover and all the other gismos necessary for supporting a web services environment, it is often cheaper to just build a custom server-based application the old fashion way - maybe with some form of proprietary terminal server product - and not have to worry about all the infrastructure. This solution may not be as versatile, but according to my friend, "versatile" isn't what is winning the bids right now.
I'm not saying these customers have it right in the long run, and, of course, the optimum solution always depends on the size of the project and the precise specifications, but the point is, this non-competitiveness of web services hasn't received much notice in Open Source circles. The last time I went to LinuxWorld, all eggs still seemed to be in the web services basket.
I wish I could tell you the custom solutions that have been winning these bids were all Linux-based solutions, but I fear that all too many of them were based on Windows. The sheer number of Windows programmers, and the familiarity of end users with the Windows user interface, give Windows a big head start. Or do they?
When you think about it, Unix-based systems like Linux are right at home in the terminal server environment, and they may have their own advantages in this market - if the gurus would put away the slide shows and start running the numbers.