The general idea is to have a look at the R implementation of “computing near the data”. In many practical applications, data is stored outside the system, in an external database. It is often possible to achieve computational cost reduction by exploiting the structure of the problem at hand: instead of downloading all the data into memory (time- and memory-consuming activity), certain elementary operations can be delegated to the database, with processed data being downloaded into the memory of an R session. Examples of this approach will be presented, along with speed benchmarks.