One of R's classic weaknesses is its difficulty in handling very large datasets, which is because R, by default, handles data by loading the full datasets in memory.

Using data analysis tools designed for large datasets, such as CERN's Root (available online at http://root.cern.ch), is one obvious solution to this problem. Root is a completely different data analysis software, and it is not easy to switch to a new data analysis platform if one has already built tools for another environment.

Some third-party R builds, including Revolution R or Renjin, have been built with memory management in mind to get around this problem. Revolution has the disadvantage of costing money to commercial users (contact Revolution about academic use). Renjin runs on the Java virtual machine, but it has the disadvantage of not being fully compatible with all R packages. For users who routinely work with very large datasets (that is a few gigabytes) and want to do it in R, it is probably...