I actually worked as a software dev at a genetics laboratory a few years ago where there was a complete network split implemented matching your description.
All resulting research data was on the airgaped network where each terminal was heavily restricted in what it was allowed to do and by whom.
We the software devs were on the open network, with less restrictions. Due to the concern of a supply chain attack there was a heavy feel of not invented here.
Over its ~25 years of operation the lab has created immense amounts of internal software from scratch.
It was an interesting look into a small dev team that had mostly never worked anywhere else (yes, the code was often very messy), and had a xenophobic view of off-the-shelve software and solutions.
I have worked in the defence sector on similar things and agree with the not invented here thing. At the same time it breeds tools out of necessity. My only tale on tale is that I pretty much designed and built ansible in Perl in 1999.
Of course, young and inexperienced, I did not see the potential preferring mild alcoholism and socialising. Doh!
You're not the only one who built a configuration management system before Ansible; CFEngine's first release was in 1993, for example. Ansible's success was its appeal to people who did not think of themselves as traditional programmers.
All resulting research data was on the airgaped network where each terminal was heavily restricted in what it was allowed to do and by whom.
We the software devs were on the open network, with less restrictions. Due to the concern of a supply chain attack there was a heavy feel of not invented here.
Over its ~25 years of operation the lab has created immense amounts of internal software from scratch.
It was an interesting look into a small dev team that had mostly never worked anywhere else (yes, the code was often very messy), and had a xenophobic view of off-the-shelve software and solutions.