I am working for a medium sized Swiss company. We're coding our own ERP, based on a nightmare of a stack. We call it "security by confusion". An attacker would maybe find its way in, but he'll never find the way out. If he destroys 90% of our code, we'll still be up and running, because 95% of the codebase is obsolete.
There was an interview with some U.S. army rep. that was pretty famous a few years back. He said that even if hackers broke in to the U.S. systems they wouldn't achieve anything due to how old, convoluted and buggy the systems were - even we who have full access can barely get anything to work!!
Of course, that was said purposefully, and may not reflect the truth at all.
But still, the thought that that is security for some (and I am sure it is for someone, somewhere) is kind of an extension of security by obscurity that is scary.
Have coded a couple of full MRP systems for various large companies based on my own model and I would be intrigued to see how this will work. I usually include the default recommended security and DR practices and then add my own OTP-hash based layer for key authorisations. I thought I was paranoid, but your system sounds more like an end-of-the-world scenario rather than just keep the production-line going in the factory.
