Open Source is great and happy and shiny, but there’s probably a commercial load balancer or twelve in the architecture for most of those sites.

At IBM, as they continue their transformation into a consulting company, they’re embracing Linux more and more, as it’s a way to simplify their offerings — they’ve been able to get it to run on all manner of hardware platforms (and they’ve got plenty).

But when it comes to maintaining or upgrading legacy software, if it works reasonably well, it is a good idea to stick with the language and OS for as long as possible, not only to recoup the initial costs, but because rewriting from scratch introduces all sorts of risks and costs. Like my well-maintained and still nice-lookin’ 9-year-old Honda CR-V, such systems shouldn’t be replaced until they can no longer do the job.

The only downside is that all of those sites are also pretty slim on the revenue/net profit side of the balance sheet. Web 2.0’s penetration into traditional areas of economic activity barely registers. Not too many mission-critical financial, manufacturing, construction, transportation, real estate, and health care applications are chugging along on open-source platforms and frameworks.

This will probably change within the next few decades, but for the moment, the old traditionals have it made in the shade because of all the cash already blown on the legacy proprietary stuff. This is why an awful lot of businesses still end up deploying (for instance), stuff based around IBM or Microsoft platforms, even though they recognise that in the very long term, this is probably not the road to stay on.

Most of them run Fedora, but there are a few Debain and Ubuntu servers as well.