How refreshing to read of a real problem and journey to an (alternative) solution, rather than yet another general description or critique of a technology or technique. Many thanks to the author.
It’s particularly fascinating for me because I work in a different world of complex line-of-business systems, which do not often involve problems like how best to implement (and persist) a very large graph, where one has requirements for fast and concurrent access with consistency constraints. That being said I have been involved in a couple of systems that incorporated “engines” that faced some similar issues. For example, a large estimation system that maintained a complete parts list for vehicles in order to calculate the cost of repairs (including the time required for repair), another being part of the calculation engine for an actuarial application, and a complex aircraft engine management system.
Would have been nice to have read of your experiences before working on the architecture of these!