Question

嘿伙计，我在oracle上运行了以下查询。

 SELECT DISTINCT
 T_TRATAMIENTO.CampaignID AS CAMPAIGNID,
 T_TRATAMIENTO.OfferID AS OFFERID,
 T_CALENDARIO.ActualDate AS ACTUALDATE,
 count(CASE T_TRATAMIENTO.CntrlTreatmtFlag WHEN 0 THEN T_TRATAMIENTO.TreatmentSize END) as NUM_OF_OFFERS,
 count(CASE T_TRATAMIENTO.CntrlTreatmtFlag WHEN 1 THEN T_TRATAMIENTO.TreatmentSize END) as NUM_OF_OFFERS_CG,
 count (distinct (case T_TRATAMIENTO.CntrlTreatmtFlag when 0 then T_TRATAMIENTO.OfferHistoryID END)) as NUM_OFF_VERS,
 count (distinct (case T_TRATAMIENTO.CntrlTreatmtFlag when 1 then T_TRATAMIENTO.OfferHistoryID END)) as NUM_OFF_VERS_CG,
 count(distinct (CASE WHEN T_TRATAMIENTO.CntrlTreatmtFlag = 0 and T_ESTATUSCONTACTO.CountsAsContact=1 THEN T_HISTORIALCONTACTO.CustomerID END)) as UNIQUE_RECIPIENTS,
 count(distinct (CASE T_TRATAMIENTO.CntrlTreatmtFlag WHEN 1 THEN T_HISTORIALCONTACTO.CustomerID END)) as UNIQUE_RECIP_CG FROM
 T_ESTATUSCONTACTO,
 T_CALENDARIO,
 T_TRATAMIENTO
 LEFT OUTER JOIN
 T_HISTORIALCONTACTO ON
 T_TRATAMIENTO.PackageID = T_HISTORIALCONTACTO.PackageID
 WHERE
 T_HISTORIALCONTACTO.CellID = T_TRATAMIENTO.CellID
 AND
 T_HISTORIALCONTACTO.ContactStatusID = T_ESTATUSCONTACTO.ContactStatusID
 AND
 T_HISTORIALCONTACTO.DateID = T_CALENDARIO.DateID
 AND
 T_TRATAMIENTO.HasDetailHistory = 0 GROUP BY
 T_TRATAMIENTO.CampaignID,
 T_TRATAMIENTO.OfferID, T_CALENDARIO.ActualDate;

表T_HISTORIALCONTACTO有8000万条记录，仍然在增长，其他表只有不到100条记录，而且还有时间响应。也进行全面扫描。我已经实现了索引，但仍然表现出缓慢的性能。

如何调整此SQL查询？你会推荐什么？我非常感谢你的帮助。 Thanxs提前

Answer 1

首先，您不需要第一个DISTINCT，因为您在前3列（CAMPAIGNID，OFFERID，ACTUALDATE）上有分组。其次，我建议避免使用非常消耗的“笛卡尔合并连接”。所以尝试这种新的连接方法。如果仍然发生笛卡尔联接，请尝试在这些表之间进行“连接”：T_ESTATUSCONTACTO，T_CALENDARIO，T_TRATAMIENTO。目前，您只将连接分隔为T_HISTORIALCONTACTO

SELECT 
 T_TRATAMIENTO.CampaignID AS CAMPAIGNID,
 T_TRATAMIENTO.OfferID AS OFFERID,
 T_CALENDARIO.ActualDate AS ACTUALDATE,
 count(CASE T_TRATAMIENTO.CntrlTreatmtFlag WHEN 0 THEN T_TRATAMIENTO.TreatmentSize END) as NUM_OF_OFFERS,
 count(CASE T_TRATAMIENTO.CntrlTreatmtFlag WHEN 1 THEN T_TRATAMIENTO.TreatmentSize END) as NUM_OF_OFFERS_CG,
 count (distinct (case T_TRATAMIENTO.CntrlTreatmtFlag when 0 then T_TRATAMIENTO.OfferHistoryID END)) as NUM_OFF_VERS,
 count (distinct (case T_TRATAMIENTO.CntrlTreatmtFlag when 1 then T_TRATAMIENTO.OfferHistoryID END)) as NUM_OFF_VERS_CG,
 count(distinct (CASE WHEN T_TRATAMIENTO.CntrlTreatmtFlag = 0 and T_ESTATUSCONTACTO.CountsAsContact=1 THEN T_HISTORIALCONTACTO.CustomerID END)) as UNIQUE_RECIPIENTS,
 count(distinct (CASE T_TRATAMIENTO.CntrlTreatmtFlag WHEN 1 THEN T_HISTORIALCONTACTO.CustomerID END)) as UNIQUE_RECIP_CG FROM
 T_TRATAMIENTO 
 LEFT OUTER JOIN T_HISTORIALCONTACTO ON T_HISTORIALCONTACTO.CellID = T_TRATAMIENTO.CellID
 JOIN T_CALENDARIO  on T_HISTORIALCONTACTO.DateID = T_CALENDARIO.DateID
 JOIN T_ESTATUSCONTACTO on T_HISTORIALCONTACTO.ContactStatusID = T_ESTATUSCONTACTO.ContactStatusID
 WHERE
 T_TRATAMIENTO.HasDetailHistory = 0 
 GROUP BY
 T_TRATAMIENTO.CampaignID,
 T_TRATAMIENTO.OfferID, T_CALENDARIO.ActualDate;

Oracle sql调优多个计数和不同

1 个答案: