spark sql data skew

[崔苗]

Hi,
when I want to count(distinct userId) by company，I met the data skew and the 
task takes too long time，how to count distinct by keys on skew data in spark 
sql ?


thanks for any reply