Example project on how to use apache spark for map reduce problem of big data
Spark is a new technology, which eases the task of solving map reduce problem. It does not require NFS and can run on local file as well on single cluster machine.
It could also run on hdfs filesystem and is appropriate in a multi-cluster environment.
You could run its jar through a native compiled spark binary, which will run it in hdfs of nfs of some sort.