|Just so you know (in the event that I ask a question here again) those inserts would be against the same database server which supports thousands of real-time users. I was experimenting because it seemed needlessly inefficient to have a clustered char(36) as the key.
Turns out that all my questions may be moot (aside from my own learning)
For review exec sp_helpindex <table name> gives the following description
clustered, unique, primary key located on PRIMARY guid column
clustered located on PRIMARY the int column
nonclustered, unique, primary key located on PRIMARY guid column
There is no detectable difference on insert speed when table size is <100k.
There is a difference, but it seems trivial, when the table contains more than two million rows (14 second difference when inserting 300k rows in each table)
Love the magnetic tape reference.
> a clustered table with 300 columns and a semi-random natural key