300 Big Data Computing
However, this comes at a cost: they may be more exible, but the interfaces
are also signicantly more low level and difcult to understand. Another
difference between MapReduce and this model is the hardware for which
each of these platforms has been designed. MapReduce is supposed to work
on commodity hardware, while interfaces such as OpenMP are only efcient
in shared-memory multiprocessor platforms.
Examples and Uses of MapReduce
MapReduce is currently being used for many different kinds of applica-
tions, from very simple helper tools that are part of a larger environment, to
more complete and complex programs that may involve multiple, chained
MapReduce executions.
This section includes a description