Just a few months ago, we passed a mainframe anniversary, of sorts. Sometime in March 1996, the IBM mainframe was supposed to be finished in the world of computing. Its end was pronounced– or rather predicted – and a precise date was given. Stewart Alsop, a head technology writer and editor with InfoWorld magazine, announced at a conference that the “last mainframe will be unplugged on March 15, 1996.” That would have been just over 25 years ago. He made that prediction sometime in 1991, about 28 years ago. Famously, Mr. Alsop ate his words in a photo he took himself in 2002.
Mr. Alsop not the only mainframe naysayer; he was just one of dozens (if not hundreds) predicting the end of the platform. And many more who profess profound astonishment that anyone would still be using such an archaic platform. “They must be stupid,” they say, or “they’re afraid to embrace new technology,” or better still, “they’re dinosaurs.”
Then there are the folks who look at the mainframe, and its current usersand managers, as obvious targets for their latest products. You know – the products that will easily translate all of that old code into something more modern, and just jolly-well run it on a few beefy AWS servers. They are the loudest detractors, that is, until you ask them the tough questions, like “why are you comparing the lowest performing mainframe hardware against the most powerful distributed servers?” Or “will your servers have the transactional throughput of the systems they’re supposed to replace?”
I often like to cite the failed science experiments using some of these “migration” or “modernization” solutions – the infamously well-known (or, rather, embarrassingly concealed) – those classic failed attempts to dump the costly mainframe – see Reboot Hill Revisited.
I think we can all agree, that the death of the mainframe is still something we can’t see on the horizon, especially considering the constant updates to the platform being made by IBM, and the large mainframe ecosystem that the platform supports. Can you see another 25 years of mainframe computing?
Well, I wouldn’t bet against it…