Wondrous Oddities: R's function call semantics [2006]

It is great for drawing vector graphics. It also has capabilities for working with large data sets, as long as you don't do it the naive way (by reading a huge array into memory).

It has very good support for sparse matrices (package Matrix) and also for serializing big data structures and loading parts in memory sequentially (like SAS). See packages bigMemory, and ff.

Actually, this view is probably the best entry point: http://cran.r-project.org/web/views/HighPerformanceComputing...

Where it's not so strong, of course, is in terms of web frameworks. Nothing that I could find, and, considering the scope of the language, it's allmost unreasonable to expect it.