I was speaking recently with the CIO of a multinational corporation about technology. We were discussing his plans to integrate technologies across some 150-plus global offices. One thing he said really struck home. He told me in his more than 20 years in IT, he never had experienced a failure on an AS/400. Sure, there was hardware that had to be replaced and the occasional software glitch, but in his career, he never lost any data because of a catastrophic failure on a midrange system.

Based on that, he said he intended to keep his data on iSeries machines as long as he could. Furthermore, he was developing a Windows-based thin client dashboard for the global enterprise built on Microsoft Office SharePoint Server 2007. That conscious decision to marry IBM and Microsoft technologies for global IT made me rethink some of my earlier opinions about the future of mainframes. Just last month in these pages I spoke somewhat disparagingly about the future of mainframe technology in an Internet world.

It also made me think about where we would be if the 1985 agreement between IBM and Microsoft had worked out a little better. The result of that joint venture, OS/2, was intended as a successor/replacement for DOS and Windows. The name came from IBM, which intended to launch the operating system on its new PS/2 line of personal computers. Both failed miserably, and Microsoft withdrew from the relationship in the early '90s. OS/2 gained some small success among anti-Redmond hobbyists and in voice-mail, telecom, and ATM systems. I actually have acquaintances who insist on using OS/2 Warp to operate their PCs. You notice I didn't say friends. I can tolerate only a certain amount of irrationality in my inner circle.