Why can't Big Blue be like Hewlett-Packard, which has driven its HP 3000 customers crazy by trying for years to kill off their minicomputers? HP stopped selling the machine in 2003 and has attempted to drive a stake through its heart ever since. Sure, those users have managed to pressure HP into extending some level of support until the end of 2010. But they're living under a death sentence, and they know it.

Like IBM, those users don't get it. We in IT have a blueprint, a road map, a grand plan. It's based on best practices, industry standards and everything else that will make the IT department look slick, smart and visionary — especially in the eyes of IT industry deep-thinkers.

Keeping legacy applications alive just because they're crucial to the business? Keeping legacy minicomputers going just because that's the only way to run those legacy apps? What kind of IT best practice is that?

Sometimes I think the crazy to get off of 'old' languages is a combination of two things - a desire not to have to learn new things and a desire to have a job. Certainly there may be many advantages to rewriting something written in assembler or (shudder) RPG in a snazzy new language. If you have a critical business system written in unstructured BASIC, by all means dump it. But, as Joel is fond of pointing out, code doesn't rust. In general, it gets better with time. If it doesn't, you can fix it. No need to start from scratch.

Labels: computers, programming, work