wrote
rbowman wrote

There must still be money in big iron which
is where IBM always was most comfortable.

I worked with a guy from Boulder in the early 90s and he said
mainframes were going to become file servers, nothing more
nothing less. That is pretty much what IBM sells now.

That's not accurate.

We call it the cloud these days but it is pretty much what the
Boulder, Raleigh, Atlanta and the other big IBM data hubs were then.

But there are far more file servers that have nothing to do with IBM.

Once we had PCs on our desk I found it a lot easier to log onto
VM, use SQL/DB2 to cherry pick a dataset, download it and use
dBase to actually crunch my numbers. It seemed faster even on
a 20m 386 and the dBase language was more powerful than SQL.

Sure, but that's not how the massive transaction processing systems are
done and they are what does the bulk of commercial computing now.

I was running pure DOS on an OS/2 (hardware) machine
so I had 12 meg of RAM to spare. I could load my whole
dataset in RAM and crunch it there. That really screams.