Microsoft's latest Windows software this year isn't all about the desktop, it is a new version of Windows Server 2003 that will be the first in several years. While it is conventional wisdom that Linux is its biggest competitor, my thought for today is that Linux will have to take a second seat to the real challenge: getting people to convert from NT. Surprise: Microsoft's secret enemy is inertia, or itself.
At our XChange conference last week, a representative from Redmond stated that there are at least 4 million NT servers out in the wild that are ripe for the upgrading. (And that isn't counting all the ones on Microsoft's own network either. Or maybe it is.) Our own surveys done for the upcoming State of Enterprise Spending issue at VARBusiness show that enterprises are still moderately investing in NT, holding about the same market share as Linux at 37 percent. No matter whose numbers you trust, that's a lot of NT lying around that no one is too thrilled about having to touch going forward.
There are several problems with an NT to WS03 upgrade. First, even with this new investment in NT (at least according to our respondents), much of the extant NT is running on hardware that is too crusty to run anything else, even Windows 2000 and especially not WS03. Corporations can't afford to refresh this hardware right now, and indeed if they are going to upgrade the software on these boxes they might as well consider moving to Linux, which isn't as resource intensive and will probably run faster on aging Pentiums than any new Microsoft OS. This is especially true as more enterprises gain Linux and Unix expertise and realize what it takes to maintain these servers. By trying to play the upgrade card, Microsoft might be encouraging more people to fall away from Windows. Now wouldn't that be a delicious irony?
Second, many of the NT crowd is still risk-adverse when it comes to running with Active Directory. Microsoft will make AD more compelling with WS03, and hopefully easier to install and configure, but I can't tell you for sure until I spend more time with the software. But again, it almost doesn't matter what the new version of AD can or cannot do, because a fair proportion of the NT crowd isn't interested in taking advantage of AD. That's because AD isn't needed: most of these NT servers are running with minimal AD installations because they are being used to host applications, either web or database servers or both, and AD doesn't really play a part here. That might change as WS03 takes hold going forward, but it will change slowly and will require a great deal of planning, forethought, and consulting effort to implement AD properly. Meanwhile, Microsoft has to spin up its AD c. 2003 training and marketing efforts, and get people convinced that the new vintage is worth paying attention to. Maybe this effort will help Novell sell more of its directory product, now wouldn't that be another delicious irony too?
Third, NT has spawned a cottage industry of its own. If properly maintained, NT can run reliably and adequately for many years going forward. I am sure that WS03 is faster and more reliable, just knowing how Microsoft works and how much time they have invested in the code to make it better. But it doesn't matter, because the old adage of not fixing something that isn't broken applies here. After all, some NT installations are coming up on their ten-year anniversaries -- that is a long time for an operating system. But it also is a testimonial to how well NT works. Take my own domain, strom.com. It has been hosted on NT 4 (running IIS v4) for quite some time. My ISP, Runtime Technologies, knows how to tweak NT to provide just enough web services and nothing more: everything else that could cause a problem has been stripped from the machine. They have had to learn this, because the early versions of NT had lots of problems. I remember hearing stories about enterprises scheduling periodic reboots of their NT servers just to keep them running more reliably. That's its own delicious irony, but I'll move on.
Fourth, the whole .Net deal is going to take a great deal of cash and effort to pull off properly. And many developers are weighing this with doing their own open systems shuffle. They are finding out that the payment to implement an all-Microsoft solution is too pricey for these penurious times. If a consultant can deliver the same application for $50,000 less by using Linux and open systems tools, they will do it. Yes, .Net offers the ability to embrace some open systems, but it is more of a casual date than going steady with a committed relationship. Indeed, the whole .Net marketing effort might do more to convince people to move towards web services and subsequently open systems than anything that the Linux crowd could ever accomplish. More irony here: we are almost having a feast.
All this adds up to the fact that Microsoft will have their work cut out for themselves going forward with WS03. And while Linux presents an interesting target, the real problem is getting people to move off of NT and to take the plunge with an upgrade. It won't be easy.