Virtual data warehouses, also called multicluster compute, are a first-order named object for a group (or cluster) of servers with multicore/hyperthreading CPU(s), memory, and temporary storage (SSD). Warehouses are massively parallel processing (MPP) engines.
Any running warehouse consumes compute credits; therefore, you should pay careful attention to right-sizing warehouses and their configuration. Not all queries require a running warehouse as some resolve their results directly from metadata or query cache, but for ...
