That should be an identity pass through or NOOP that does nothing, but MapReduce requires "something".
On Fri, Jan 18, 2013 at 8:57 AM, nagarjuna kanamarlapudi < nagarjuna.kanamarlap...@gmail.com> wrote: > No but the query execution shows a reducer running .. And infant I feel > that reduce phase can be there > > > On Friday, January 18, 2013, Dean Wampler wrote: > >> There is no reduce phase needed in this query. >> >> On Fri, Jan 18, 2013 at 6:59 AM, nagarjuna kanamarlapudi < >> nagarjuna.kanamarlap...@gmail.com> wrote: >> >>> Hi, >>> >>> Select col1,myudf(col2,col3) from table1; >>> >>> >>> In what phase if map reduce an udf is executed. >>> >>> In the very beginning, I assumed that hive will be joining two tables., >>> getting the required columns and then applies udf on columns specified >>> I.e., essentially on reducer phase . But later on I realised that I was >>> wrong. >>> >>> Is there any specific parameter which suggests hive to call udf at >>> reducer phase rather than at Mapper phase. >>> >>> >>> Regards, >>> Nagarjuna >>> >>> >>> -- >>> Sent from iPhone >>> >> >> >> >> -- >> *Dean Wampler, Ph.D.* >> thinkbiganalytics.com >> +1-312-339-1330 >> >> > > -- > Sent from iPhone > -- *Dean Wampler, Ph.D.* thinkbiganalytics.com +1-312-339-1330
