MapReduce offers a robust and conceptually simple framework for imple-menting index construction in a distributed environment. By providing a semiautomatic method for splitting index construction into smaller tasks, it
can scale to almost arbitrarily large collections, given computer clusters of sufficient size.