求助: 面對(duì) 千萬級(jí)別的數(shù)據(jù) 進(jìn)行統(tǒng)計(jì),誰有解決方案? 過程:一個(gè)表,表有70列,3千萬行, 先用where篩選 可能 還有 5個(gè)左右like查詢,group by 三個(gè)列,得到新的table,再對(duì)這個(gè)table 進(jìn)行至少2層嵌套的sum,count等.求答案.先謝謝了.1)原表3千萬行, where篩選出來的就有500萬行~800W.再對(duì)這個(gè)結(jié)果進(jìn)行至少2次嵌套統(tǒng)計(jì),  中間統(tǒng)計(jì)我想過了,但因?yàn)樘峤坏臄?shù)據(jù)是有范圍的,即范圍和條件不固定.無規(guī)律可找,所以沒法提前處理數(shù)據(jù),只有等客戶提交條件后才能方向統(tǒng)計(jì)2)下面是一條sqlexec('SELECT CASE WHEN 參與評(píng)價(jià)處方數(shù) IS NULL THEN 0 Case這里省略10個(gè)Caseinto ['+@OldTable+']from(SELECTCOUNT(1) as 參與評(píng)價(jià)處方數(shù),SUM(YPZS) as 用藥品種總數(shù),這里省掉10個(gè)以上sumSUM(JZKJYCFCS) AS 急診患者抗菌藥處方數(shù)FROM(SELECT VISIT_DATE AS CFRQ, PATIENT_ID AS BRBS, ORDERED_BY AS KSDM,COUNT(1) YPZS,SUM(CASE WHEN DNDCID IS NOT NULL THEN 1 ELSE 0 END) AS TYMCS,下面省掉10個(gè)以上sumAS JZKJYCFCSFROM{0}GROUP BY VISIT_DATE, PATIENT_ID, ORDERED_BYHaving 1=1 {1}) AS TEMP ) as ACREATE TABLE ['+@Table+']這里省掉70個(gè)列,下面是建表和索引.針對(duì)新查出來的表UPDATE ['+@Table+']SET 平均每張?zhí)幏接盟幤贩N數(shù) = case when下面再省掉所有列的update部分,下面是 row_number()的分頁3)以上語句大寫字母是列的名稱或別名,{}里面是傳入的其它條件或表名