select b.* from (select a.*,row_number() over (partition by 列1 order by 列2 desc) rn from a) b where rn=1; 列1 是分组的列