When you install and configure SAS Analytics Pro on a Windows Docker Desktop, issues might occur where performance is catastrophic or your storage keeps growing in size. This SAS KB article explains how to resolve both of these scenarios.
The following sample SAS code creates about 32GB of data in the Work library.
data test;
set sashelp.cars;
do i=1 to 500000;
output;
end;
run;
If there is only one active user, problems do not occur. However, issues occur once other users begin to run code that creates large Work library tables. Suddenly, Docker consumes a lot of storage in a location similar to the following: C:\Users\USERNAME\AppData\Local\Docker\wsl\disk\docker_data.vhdx.
This issue occurs because the storage consumed by docker_data.vhdx never decreases. In SAS Analytics Pro, the Work library clears when you restart the container. However, the size of docker_data.vhdx remains the same even after you restart the container.
To work around the storage issue, mount the Work library outside of the container. See Set New Locations for the Work Library and Utility Files for instructions.
You can remove the "--detach" option to better troubleshoot the launch script.
With this workaround, the Work library no longer uses the storage within Docker. Instead, the Work library uses the host storage.
Note that, completing this workaround introduces performance issues, as described in the next section of this article.
The Storage Workaround introduces a new issue. After completing the Storage Workaround, running the sample SAS code takes far more time to complete than it initially did. For example, in a Test environment, the code took more than ten times longer to run than it initially did.
This new issue occurs when the Work library is mounted from the Windows OS, usually with a C drive or a D drive. When Docker tries to run an I/O heavy application like SAS, it has to go through the Windows-Linux boundary, which creates a huge bottleneck.