As I begin to execute my team's initiatives for 2005, I'm finding that open source is still central to my overall IT strategy. My view of open source in the enterprise, however, is changing as open source matures.
When I first rolled Linux systems into production seven years ago, I have to admit that the "free" part of open source was the main attraction - "free," as in beer, as opposed to the more abstract "free," as in speech (see the GNU free software definition).
Of course, I was impressed with the stability and overall capabilities of Linux even then, and I admired the revolutionaries who created such great new software.
In reality, however, I was not so much an ideologue as an IT manager with racks full of Intel hardware, a long list of problems to solve, and little cash to spend on OS licenses. Linux was perfect for that environment.
Fast forward to 2005 and open source means different things than it did back then. Open source solutions have clearly moved beyond the initial "free, as in beer," appeal. According to IDC predictions for 2005, Linux shipments will account for more than 20 percent of server volume shipments, growing at twice the rate of Windows. Most interestingly, IDC notes that within the manufacturing, financial services, telecom, and government verticals, organisations are clearly moving toward enterprise-grade, commercially supported Linux distributions - that means "paid." I, too, am now willing to pay for the functionality I have enjoyed at little cost for years now.
In enterprise IT, the days of downloading a Linux distribution and compiling it yourself are mostly over - although you can still do it. Linux and open source in general remain compelling because they form the basis of a solid technology stack for many applications. In the late '90s, IT was getting good open source software such as Linux for free. Even though we're now starting to pay a little for open source software, it's still a great bargain. I'll definitely be watching companies such as SpikeSource that are moving beyond supporting a base open source OS into supporting an integrated and vigorously tested enterprise stack of open source applications. If successful, this type of support might just make building on open source a no-brainer.
I continue to be excited about open source solutions, both free and paid, but I temper my breathlessness with a dose of caution. Open source solutions are not automatically superior to commercial or outsourced approaches, and what works in one context might be burdensome in another.
Last year, the InfoWorld IT team implemented SpamAssassin, an excellent open source spam-filtering solution, and it works well most of the time; however, because InfoWorld is a technology media company and our writers' and editors' emails are published all over the site every day, I think we get a larger proportion of incoming spam than the average IT operation. Sometimes our multiple SpamAssassin boxes with fast processors don't keep up as well as I would like. We could easily keep throwing hardware at the problem, but that increases the operational burden on a lean staff. So, we'll probably outsource the problem altogether.
Although open source doesn't solve every problem, 2005 might be the year when open source infiltrates the enterprise IT mainstream. The free ride might be over, but a few modest checks are a small price to pay for enterprise-class software.