"the bank has still has systems from mergers that were 20-30 years ago! I assume they've been updated a bit in that time"
Erm... My guess is it probably went something like this...
- immediately after the mergers: the quickest way to get stuff running is to build a quick-and-dirty interface to pipe data between systems. What we can't match automatically can be reentered manually. We'll migrate the systems properly later.
- +3 years: We hadn't realised before, but there are fundamental differences between data models of the various systems, which make the 'proper' fix quite expensive
- +6 years: Management starts to finally overcome the cost resistance and reluctantly start to explore the possibilities of a proper fix
- +9 years: Studies are in, there is management buy-in and budget available, but all the key people have retired or left and hardly anyone remains who understands the requirements anymore
- +12 years: Ah, fuck it, it's been working for over a decade so why change? The mess will all go away once our new global banking platform is online in the next 2 years
- +15 years: The mess will all go away once our new global banking platform is online next year
- +18 years: The mess will all go away once our new global banking platform is online next year