to expand: if you have a different schema with no overlapping data, then sharing a cluster would just introduce the risk of a noisy neighbor, so we recommend a distinct cluster in that case