SQL and MapReduce, was not enough: even more different models were needed! And due to the openness of the Hadoop platform, it was feasible to implement a whole range of approaches, which would not have been possible within the confines of a monolithic MPP database
