Why is it recommended that Arcadia State Store and Arcadia Catalog Cache run on the same node in our cluster?

The job of the Arcadia Catalog Service (catalogd) is to relay the metadata changes from Arcadia SQL statements sent from the Arcadia Visualization Server (Arcadia UI) or ‘arcadia-shell’ to all the Arcadia Analytics Engine (arcengined) daemons in a cluster. However these metadata requests are passed through the Arcadia Statestore Service (statestored), which then is tasked with broadcasting the metadata to rest of the Arcadia Analytics Engine (arcengined) daemons. Therefore, because there is such frequent communication between these services, its more efficient to have them both running on the same node.