Hey Jason, While I am not sure on whats the best way to automatically "evaluate" during the execution of a job, the MultipleInputs class offers a way to run different map implementations within a single job for different input paths. You could perhaps leverage that with duplicated (or symlinked?) input paths.
Otherwise, perhaps do all the N types of computation in a single map() call, and judge the time inside it at the end of all, before emitting? On Tue, Sep 11, 2012 at 9:03 AM, Jason Yang <lin.yang.ja...@gmail.com> wrote: > Hi, all > > I've got a question about how to make different mappers execute different > processing on a same data? > > Here is my scenario: > I got to process a data, however, there multiple choices to process this > data and I have no idea which one is better, so I was thinking that maybe I > could execute multiple mappers, in which different processing solution is > applied, and eventually the best one is chosen according to some evaluation > functions. > > But I'm not sure whether this could be done in MapReduce. > > Any help would be appreciated. > > -- > YANG, Lin > -- Harsh J