hadoop - Extending Hive: writing a UDF that does both Map and Reduce operations -
i working on project extend hive support image processing functions.
to this, need read in image
, break multiple files, pass each separate map task processing on , reduce them 1 image returned user.
to this, had planned implement udf
call mapreduce
task in hadoop. however, understand udf
operate either on map side or reduce side of hql query
, while need ideally 'bridge gap
' between map , reduce side.
the hive documentation
isn't helpful, , looking pointers on start looking more information this. please feel free ask more questions if haven't been clear enough in question.
looking hipi (hadoop image processing inteface) might give start. particularly, example on computing principal components of bunch of images might of interest.
Comments
Post a Comment