Which of the following says Hadoop provides does Pig break?
(1)Calls to the Reducer's reduce() method only occur after the last Mapper has finished running.
(2)Task stragglers due to slow machines (not data skew) can be sped up through speculative execution.
(3)All values associated with a single key are processed by the same Reducer.
(4)The Combiner (if defined) may run multiple times, on the Map-side as well as the Reduce-side.
