SAS is big in the corporate world. I work in the finance industry and SAS is still quite big. To be honest, using SAS is a pain. SAS can’t even syntax highlight its own language properly. But it’s still there because it has one trick – disk-based data manipulation and associated algorithms.
Ten years ago when I introduced R to my workplace, people were skeptical – you can’t load a large dataset in R and manipulate it like in SAS. That’s because in R the dataset needs to be loaded into memory and at that time the largest laptop only had 4G of RAM. Today, 32G RAM laptops are becoming the norm but still I can’t load really large (50G) datasets into RAM.
I think Julia and R can replace SAS by implementing disk based data manipulation as a first class citizen. Also can most algorithms works once the data becomes disk based? If not then Julia still can’t replace SAS, because most of SAS’s algorithms (e.g. proc glm) works off disk based data