This project is a prototype implementation of an on-disk indexing structure called buffer trees. Buffer Trees were introduced by Lars Arge in 2003. They are an extension of the traditional B-Trees which are optimized for writes. Buffer trees are known to provide guaranteed bounds on number of seeks required to access a tuple from disk for a given size of the database.
In addition to the idea of buffer-trees, this implementation includes some optimizations for primary-key indices using bloom filters.