Interesting performance issue with high cardinality indexing

It’s a bit unclear what you mean here. Are you using the @count index in your schema? Get started with Dgraph

Can you share an example of a mutation in your cluster so we can get a better idea of the structure of your graph?