Recently i came across
FIRST_DAY_OF_YEAR UDF in SQL which returns Jan 1st of that year.
This could be made simply using existing functions in Hive as
suppose our input date is '2016-03-04'
instead of loading a JAR file. And moreover i guess using JAR has less performance when compared to the above implementation.
We can also use UDFs for hiding complex functions from our query which makes it more readable.
Kindly let me know if i'm wrong about performance.