A virtual warehouse is basically a collection of computing resources (like CPU, memory, Solid state drive, etc.) that customers can access to run queries, load data, and perform other DML (Data Manipulation Language) and SQL (Structured Query Language) operations.
For example, it provides memory, temporary storage, and CPU resources that can be used for DML operations and SQL execution. You can use this independent compute cluster at any time and then turn it off when not needed. You are charged (paid) for each virtual warehouse you run, their size, and how long they run. Virtual warehouses do not share their compute resources with each other, and therefore, their performance is independent of each other.
As shown in the following diagram, different groups of users can be assigned separate and dedicated virtual warehouses. Therefore, ETL processes can continuously load and execute complex transformation procedures on separate warehouses, ensuring no impact on data scientists or finance reports.
