How Cockroach avoids Cross-node Joins?

Hey Folks,

I have watched a few talks and one thing that I’m curious about is how do you guys avoid cross node joins? Let me explain: One of the issues with a distributed relational is that you might end up with your data in multiple nodes, and once you need it back, you might need to go through many nodes to get everything you need.

How do you guys avoid, let’s say, the User Data (which normally is stored using at least +5 tables) to be spread between multiple nodes? is there some kind of sharding based on the data structure (not on the fields)

Best,