Missing Mainframers

Computerworld - Suppose there's a shortage of a particular IT skill set. We all know what happens, don't we? Those skills become hot. Training courses spring up. Salaries go through the roof. And the one thing that surely won't happen is that people with expertise in that skill set will be retrained to do something else, thus making the shortage worse -- right?
Except when it comes to mainframes. Crazy, huh?
As Computerworld's Patrick Thibodeau reports this week (see story, page 1), mainframers are in such short supply that one professional group for data center managers is setting up its own college-level online courses for potential mainframers. But at the same time, IT shops that are phasing out mainframes are actually retraining experienced mainframers to do nonmainframe IT work.
Insane as that sounds, it does make sense when you remember that 55% of mainframers are in their 50s or 60s. These aren't kids who are ready to jump to another company at the drop of a hat. These people are fully vested in their pension plans. They'd rather struggle to learn new technologies than start over at another company that needs their mainframe skills for the few years before they retire.
Meanwhile, young IT people aren't jumping into mainframe work. They were never taught mainframe skills. After all, pundits and analysts and other industry blowhards have been saying mainframes are dead, or at least dying, for a decade now. Why learn something that's dying?
But mainframes aren't dead at many IT shops. And they show no signs of dying soon. Unix didn't kill them. PC networks didn't kill them. Y2k was supposed to kill them when all that mainframe code stopped running on Jan. 1, 2000. But instead we spent billions of dollars fixing that code -- and now we have to earn out that investment.
That's not the only reason mainframes refuse to die. Mainframes work. They still set the standards for security and reliability and availability in IT. And mainframe MIPS have kept getting cheaper in the 10 years since IBM realized it could no longer survive by soaking its mainframe customers.
So mainframes aren't going away. But mainframers are. And they're not being replaced.
What do you do if you're looking at a staff of highly experienced mainframers who will be retiring in a few years? You can outsource, but that limits your flexibility -- and it hands the management of your most critical data to someone for whom it's just another job. You can migrate off those mainframes,