Not entirely correct. The distributed database can scale computation horizontally across the nodes in cluster. Single node is better only for the relatively sm

Not entirely correct. The distributed database can scale computation horizontally across the nodes in cluster. Single node is better only for the relatively small dataset. If you go higher volumes and velocity, Vertical scaling is still more expensive than Horizontal scaling. If your data is large, with network bandwidth of 1Gbps (or perhaps better) distributed database may choose to replicate it efficiently across the nodes and do scatter-gather (or divide and conquer) query parallelization and also maintain HA. Citus DB is one example of distributed database which does exactly that.