I mentioned in my last post in this series, that the Compute layer of Snowflake is basically made up of a series of Virtual Warehouses (VWs). Each VW is an MPP (massively parallel processing) compute cluster that can comprise one or more compute nodes.
The number of nodes in the compute cluster is called its "size" and the sizing options are made to resemble T-Shirt sizing, as you can see in the main image above.
Note that XS (extra small) is the smallest VW size. Using an XS for one hour consumes one Snowflake credit. In Australia, in Azure, right now, that's a little over $2 USD.
The number of credits used per hour for a VW is directly related to the number of compute nodes that it contains. So a 2XL VW consumes 32 credits per hour.
The other important aspect of this sizing is the number of threads. Currently, there are 8 threads per compute node.
So the number of concurrent threads for a VW goes from 8 at the XS level, up to 1024 at the 4XL level.
I'll talk more about threads later but they become important when you're trying to get good parallel loading of files happening, and when you want significant concurrency in query execution.
For an index to all posts in this series, see the first post here.