Is it fair to assume a single MV is going to be maintained by a single RW compute node at most?
Nizar is asking about the maintenance of materialized views in a distributed system and whether they are handled by a single compute node or multiple nodes.