Let's assume I have 3 machines with me.
My first machine has SQL Server 2016 with R Services installed on it.
My second machine has SQL Server 2016 without R services enabled on it.
My third machine has a standalone Microsoft R Server installed on it.
I understand that since my first machine has SQL Server with R Services, I can store R script as stored procedures on SQL Server and call it later for analysis.
My question lies with how my standalone Microsoft R Server connects to either SQL Servers (with and without R Services enabled).
From what I understand, it uses ODBC to connect to SQL Server and uses the RevoscaleR functions to query or analyze the SQL data - in both cases.
If that's the case,
- What benefit or limitation does my machine 3 (standalone MRS) have while connecting to SQL Server with R Services vs. SQL Server without R Services?
- In either cases, can I invoke any stored procedures from my R IDE remotely?
- Can I push the compute context to SQL server in both cases?
- Is there anything special that SQL Server with R Services offers for a standalone Microsoft R Server when compared to a SQL Server without R Services.
- Is there any difference in the database engine's ability to perform other tasks either way?
Imagine I'm reluctant to enable R Services on my SQL Server because it may take up the memory and slow down the performance of the database engine itself. All I want to do is get a standalone Microsoft R Server and connect it to SQL Server without R Services and get all the scalability and performance anyway.
What is it that I'm missing by doing this?