An article in Computer world magazine reports on Air Force scraps massive ERP project after racking up $1 billion in costs. The system was meant to replace some 200 legacy systems.
According to the article, Oracle ERP was supposed to be used to replace the 200 Legacy systems containing business logic which have evolved over many years. I have no doubt that Oracle ERP is a very good and competent system - and most of the difficulty would have been the merge of the business logic from 200 dispersed systems.
I wonder what went wrong. Was it difficult to extract the business rules, was merging them the issue, was the system simply too big for anyone to master ???
Manual rewrite of legacy system is a large software project, the kind which have a tendency to go over time and budget. Hence the argue goes that an automated translation, (one which can generate high quality translation) is a lot less risky than a "manual project", and have a better chance of completing within budget and on time.
But this time, they weren't rewriting the whole thing, but using an established ERP system. I guess the same argument applies.