The two approaches are essentially them same, although in 2.1 the meta
node benefits are built into the store format.
If you're querying relationships that are in a minority group
(type+direction) there's a good benefit, but if you query relationships in
a majority group, i.e. the most common
In my case, as the edges of a node grow the performance becomes worse.
My use case includes traversing all neighbor nodes and their neighbor
nodes. But the problem is that their relations of two types are growing.
I've read the following board.