I'm pretty sure you can get comparable performance from a graph in SQL if you make sure you have covering indexes for the Node ids and the edges/adjacency list

I'm pretty sure you can get comparable performance from a graph in SQL if you make sure you have covering indexes for the Node ids and the edges/adjacency list table. This may be implementation-dependent but in theory, it's solid. I'm not sure what else a graph database would be able to do to beat an indexed adjacency list.