Are there some case studies, lessons learned, best practices around transformation of large legacy platforms based on COBOL and monoliths?
The platform in question is highly available, executes on a single node, and works well in terms of availability - a couple of hundred transactions per second at peak load. Business rules are embedded across code modules built over three decades. Parts of the database are not quite SQL, but close. Changes to certain business rules take months if not years, and have to be planned accordingly.
There are several systems which feed into or out of it, also aged (but not COBOL). The core system critically depends on feed from some of these associated systems, while other systems not part of the core would stop functioning if the feed to them were to go down more than an hour or a few hours. Some of the associated external systems are also business critical, and need to be highly available.
The obvious transformation is of course to build a system in parallel using a modern evolutionary architecture, chopping up the current monolith. What I am looking for is the next level of detail - case studies, lessons learned, best practices, and whether there is are well known (based on actual transition) set of best practices to be followed, whether frameworks have been developed etc. Kindly note: this is not about "How to get off COBOL", or "Why do you want to change / Why do you want to get off COBOL", "Did you know that COBOL is making a comeback" etc. It's just about the detailed case studies - if they exist - on modernizing to a new platform without taking the risk of a big bang transition.