Client-Side R Versus Server-Side R

IML Studio and the SAS Workspace Server have a client-server relationship: IML Studio runs on a desktop computer (the client) and the SAS Workspace Server runs on a server computer (the server). It is possible, of course, for IML Studio and the SAS Workspace Server to run on the same computer.

Before you can use R with IML Studio, you must decide on which computer you want to run R. You have the following choices:

• Run R on the client computer. This configuration, which is referred to as client-side R, is the recommended configuration.
• Run R on the server computer. This configuration is referred to as server-side R.
• Run R on both the client computer and the server computer.

It is recommended that you use client-side R with IML Studio unless you have a specific need to use server-side R. The reasons for this advice are as follows:

• The interface between IMLPlus and client-side R is more extensive than the interface between IMLPlus and server-side R.
• With client-side R, you can stop a long running computation or infinite loop in your R code. IML Studio does not enable you to stop execution within server-side R.
• The interface to client-side R does not require you to add the RLANG system option to the SAS Workspace Server configuration file.
• At most sites, it is much easier to install R and R packages on your computer than to get a system administrator to install R and R packages on a SAS server.
• With client-side R, there are no additional security considerations because R runs under the same user account as IML Studio. With server-side R, there are additional security considerations because R runs in the security context of the SAS server. To use server-side R, the RLANG system option must be enabled. The RLANG system option has the same security implications as the XCMD system option.
• With client-side R, each user of IML Studio can potentially use a different version of R with a different set of R packages. With server-side R, everyone who connects to the server must use the same version of R and the same set of R packages.