Hi @benoitc,
We’d still use the same ports, same RocksDB directories etc. But, we’d make it so that the data is internally segregated; all the queries would be limited to that silo; with no chance of mixup.
That way, we can support multiple different clients on the same instance of Dgraph.