I worked at a startup (now fairly popular in the US) where we had tables for each thing (users, companies, etc) and a “relationship” table that described each relationship between things. There were no foreign keys, so making changes were pretty cheap. It was actually pretty ingenious (the two guys who came up with the schema went on to get paid to work on k8s).
It was super handy to simply query that table to debug things, since by merely looking for a user, you’d discover everything. If Mongo was more mature and scalable back then (2012ish), I wonder if we would have used it.
It was super handy to simply query that table to debug things, since by merely looking for a user, you’d discover everything. If Mongo was more mature and scalable back then (2012ish), I wonder if we would have used it.