[kephra]

I would advise to hands on machine. You can run many classical machines in emulators like Hercules now.

e.g. Try a bit on COBOL and REXX to process records in a batch. Next play around with Panel Features, Overlay Generation, CICS and CODASYL to write classical interactive applications. This will show that web is just 3270 on steroids, with a much bigger memory and CPU footprint.

Now read Stonebraker and C.J.Date, and realize what a relieve relational medium size databases had been. That the burden of non relational databases only makes sense for small size high performance databases that fit into RAM, or big databases that have to scale over a cluster.

You can also explore Lisp, Smalltalk, Forth, and P-Code machines. Operating Systems like Oberon or Spin. Each of them will offer an insight of how virtual machines evolved. And languages like Lisp, Smalltalk, and Forth are eye opening as they enforce their paradigm.

[brudgers]

For me, understanding the history of computing goes hand in hand with a willingness to engage any computer programming language. It is easy not to learn that COBOL solved really hard problems of data storage and control flow and organizing teams of programmers against big big projects in the days before...well before just about everything we see in computing today. COBOL was NoSQL by necessity.