Can reverse edges be added later to a production database without deleting or effecting existing data?
If it does effect the data, does adding reverse edges to all edges right from beginning (even if you don’t use them right away) consume a lot of resources (memory, etc)? I assume the worst case is that they take as much resources as any other edge unless the system somehow treats them differently and is able to optimize.
It’s usually recommended to always include reverse edge from the beginning, unless memory/disk is an absolute concern. Because schema change doesn’t support adding reverse edge, the edge has to be reloaded to add reverse edge. And query optimizer will take advantage of reverse edge to traverse both directions, which can significantly improve performance. Memory/disk wise it will double the consumption for the edge type, but the TG memory footprint is already highly optimized so it’s usually not a concern
That is an helpful answer. Thx!
