The last blog post in the Microsoft SQL Server R Services series covered some performance related parameters of sp_execute_external_script, (@parallel and @r_rowsPerRead), and it touched a little bit on SQL Server Compute Context.
After I published the post, I received an email from Umachandar Jayachandran who is a Principal Program Manager in the SQL Server team and is working on integration of R, Python and other runtimes into SQL Server (basically, he knows this “stuff”). In the mail, he said that I was somewhat incorrect in what I said about how the SQL Server Compute Context (SQLCC) works.
So I decided to write a post to correct my mistakes, and this is it. While I wrote this post I realized that there are more to the SQL compute context that meets ones eye, so I will - in all likelihood - write more posts about SQLCC.
To see other posts (including this) in the series, go to sp_execute_external_script and SQL Server Compute Context.
Before I begin I want to say a very big THANK YOU to Umachandar as well as Nellie Gustafsson who also is a SQL Server Program Manager at Microsoft. Without them making me see the “errors of my way” I would not have been able to write this post.