How would a N1QL join statement behave in a sharded environment?

How would it be executed so that it’s performant?

N1QL joins are designed to be performant in a sharded environment. For you as a developer, you only need to ensure that your queries are using an index. Use the CREATE INDEX statement to create an index on your filter field(s), and use the EXPLAIN statement to verify that your queries are using an index.