Building something new from scratch is so much easier than modifying existing systems. You don't have to worry about anything yet. No legacy code or undocumented decisions from the past, no outdated protocols, no cable harnesses that nobody dares to touch. You simply choose the best components available today and build it the way it should be.
You don't have that luxury when you need to modernize an existing system. Imagine you have a system or installation that has been running well for years. Parts are becoming harder to source or have reached end-of-life. The software runs on an operating system that is no longer supported. Customers are asking for remote monitoring, dashboards, and integrations with their own systems. The market expects functionality that simply didn't exist ten or twenty years ago.
But now you have to upgrade these existing systems while everything keeps running. Maybe even without users noticing. How do you approach that? Step by step, or big bang all at once? Do you even still know how all subsystems work and communicate with each other? Is there documentation, or does the knowledge live with that one colleague who's been there for twenty years? Can you run the new alongside the old, or do you have to flip the switch on a Sunday night and hope everything works? And how do you even test that?
Does this sound familiar? I'll be diving deeper into this topic in the coming weeks. But I'm also curious how you handle this. Incremental, big bang, or just postpone as long as possible?