java - spark sql对单表数据在1.6亿左右的表进行复杂查询，速度有多快

Question

sql大致如下
select t.numIid numIid,t.keyCount as keyCount,t.ssrqCount as ssrqCount,

        ((t.keyCount-t2.keyCount)/t.keywordCount) as growthIndex,t.keyCount/#{KeyCont} as keyPercent
        from
        (
            select numIid, 
            sum(keyCount) as keyCount,
            sum(ssrqCount) as ssrqCount
            from
            tablename
            where
            day="201601"
            and type in(1,2,3)
            group by numIid
        ) t
        left join
        (
            select numIid, 
            sum(keywordCount) as keywordCount,
            sum(ssrqCount) as ssrqCount
            from tablename
            where
            day="201602"
            and type in(1,2,3)
            group by numIid
        ) t2
        on t.numIid =t2.numIid
    order by keycount desc